Revisiting the relative effectiveness of slaughterhouses in Ireland to detect tuberculosis lesions in cattle (2014-2018).

Slaughterhouse or meat factory surveillance to detect factory lesions (FL) at slaughter is an important part of the bovine tuberculosis (bTB) eradication program in Ireland. The objective of this study was to quantify the effectiveness of Irish slaughterhouses or factories in submitting FL and the proportion of those submitted FL confirmed as being due to bTB in slaughtered cattle, and to identify and quantify the association of risk factors at animal, herd, and factory level with FL submission and confirmation. The data consisted of 6,611,854 animals slaughtered in Irish factories from 2014 to 2018 obtained from the Department of Agriculture, Food and Marine (DAFM), Ireland. Selected risk factors for this study included factory, year and month of slaughter, age, sex, breed, animal movement, ever inconclusive in the standard or severe skin test, herd type, herd size, and bTB history. The association of each risk factor on the FL submission and confirmation risk were analysed with univariable followed by a multivariable logistic regression with herd as random effect. Factories were ranked and compared based on the odds ratio (OR) obtained from the univariable (crude OR) and multivariable (adjusted OR) analysis. The average submission risk of all factories was 20 per 10,000 animals slaughtered, ranging from 1 to 42 per 10,000 animals slaughtered, and the average confirmation risk over all factories was 40.72%, ranging from 0.00 to 61.84%. The odds of submitting and confirming FL as bTB positive were higher in animals over eight years old compared to animals 1-2 years old (OR = 1.91, 95 CI 95% 1.77-2.06 and OR = 4.05, 95% CI 3.17-5.18, respectively), and were higher in animals that ever had inconclusive skin result based on severe interpretation (OR = 2.83, 95% CI 2.44-3.27 and OR = 4.48, 95% CI 2.66-7.54, respectively), animals originating from sucklers herds (OR = 1.08, 95% CI 1.02-1.14 and OR = 1.31, 95% CI 1.11-1.55, respectively), or herds with bTB history in the past three years (OR = 4.46, 95% CI 4.28-4.66 and OR = 319.90, 95% CI 237.98-430.04, respectively). The odds of FL submission and confirmation decreased as the herd size increased (OR = 0.95, 95% CI 0.93-0.96 and OR = 0.82, 95% CI 0.78-0.86, respectively). An inverse relationship of FL submission and confirmation was present for variable sex and inconclusive skin result with standard interpretation, where submission odds were higher in males (OR = 1.05, 95% CI 1.00-1.10) and ever inconclusive animals (OR = 74.24, 95% CI 69.39-79.43), although the confirmation odds were lower (males OR = 0.66, 95% CI 0.56-0.76; ever inconclusive OR = 0.44, 95% CI 0.36-0.54). The crude and adjusted ranking of factories did not differ greatly for FL submission, indicating that factory-related factors may contribute significantly to the submission variation between factories. However, a substantial difference between crude and adjusted confirmation ranking was present which may indicate that animal and herd-related factors were associated to variation in confirmation risk between factories.

Introduction

Bovine tuberculosis (bTB) is a chronic infectious disease caused by the Mycobacterium tuberculosis complex, especially Mycobacterium bovis, affecting cattle, wildlife, and humans worldwide. In industrialised countries, M. bovis infection has less zoonotic impact compared to developing countries [1] but it is considered of high economic importance as it causes losses at the farm level and negatively affects international trade in Europe [2]. Due to successful eradication programs, 17 of 28 European member states had gained the Official Tuberculosis Free (OTF) status by 2019 –Ireland being among the 11 non-OTF countries [3].

Ireland has been facing the challenge of eradicating bTB for a long period, and introduced a national eradication program in 1954 [4]. The first recorded data showed approximately 90,000 bTB reactors in 1959 [5] (cattle population of 4.68 million cattle; crude prevalence 1.92% [6]), a number that steadily declined to approximately 17,500 reactors by 2018 [5] (cattle population 7.34 million cattle; crude prevalence 0.24% [6]). Despite this promising reduction, bTB has not yet been fully eradicated from Ireland, which is considered to be partially attributable to the spill-over infection to and from an abundant wildlife reservoir, the European badger (Meles meles) [7] and maybe deer [8]. Recently, a rise in herd incidence of bTB infection from the lowest level of 3.27% in 2016 to 4.37% in 2020 has been observed [9]. This increasing incidence has triggered the Irish government to launch a renewed eradication strategy: ‘Bovine TB Eradication Strategy 2021–2030’ with the goal of driving down bTB levels by 2030 towards eradication [9, 10]. The program includes active surveillance with annual herd testing with the single intradermal comparative tuberculin test (SICTT), slaughterhouse (factory) surveillance at the slaughter line to detect bTB gross lesions from antemortem test negative animals, termed factory lesions (FL), as well as the removal (culling) and vaccination of badgers [11].

The annual SICTT targets all bovines over six weeks old, and combined with culling programs it aims to detect and eliminate reactor animals in the early stage of infection or preclinical stage [12]. Next to culling reactors, the herd of origin is restricted from trade and repeatedly tested at sixty-day intervals, in addition to herd-specific epidemiological investigations and supplementary testing where appropriate. If two consecutive herd tests are negative, herds regain an Official TB Freedom (OTF) status and trade restrictions will be lifted [10]. However, herds with OTF status still have the probability of having M. bovis infected animals due to the imperfect sensitivity of the SICTT [9], which has been reported as 90.6% [13] and between 52.9% to 60.8% [14] under Irish conditions using traditional and Bayesian approaches, respectively. In a meta-analysis of studies based on data from Ireland and the UK, a sensitivity of 50% (95% CrI 0.26, 0.78) and a median specificity of 100% (95% CrI 0.99, 1.00) was estimated [15]. Therefore, routine inspection by certified veterinarians for presence of bTB-like lesions in organs, head and thoracic lymph nodes of slaughtered animals is an essential part of bTB surveillance program in Ireland. When an FL is suspected to be due to M. bovis, a sample is taken and submitted to the national TB laboratory for confirmation [10]. Detection of FL has been indicated as a crucial component in the surveillance program as it can disclose bTB infections in attested cattle–i.e., in cattle that originate from OTF herds which are not thought to be infected at the time they are sent to slaughter [16]. In the period 1989–1997, slaughterhouse surveillance disclosed annually 9% to 33% of all herd breakdowns in Ireland [17], while in a recent report, 20,116 bTB breakdowns in the 2005 to 2019 period were commenced by the disclosure of factory lesions [18]. In Northern-Ireland, this percentage was between 18% and 28% [19].

Slaughterhouse surveillance to detect FL at slaughter is part of the bTB eradication program in many countries, not only in Ireland but also in Great Britain [20], Northern Ireland [19, 21], Spain [22], and the United States [23]. Slaughterhouse surveillance based on meat inspection (syn: post-mortem inspection) also has been used to investigate disease epidemiology and traceback investigations of others diseases such as fascioliasis [24] and bovine cysticercosis [25]. Detection of bTB lesions at slaughter is not only detecting part of the infected cattle that remained undetected from routine skin testing but also as a tool for early detection of infected herds [19], and as a ‘whistle-blower’ to prevent the spread of infection to other herds. Continuous and regular monitoring of animal, herd and factory-related factors through slaughterhouse surveillance is important as quality control of the bTB eradication program [26].

As slaughterhouse surveillance is an important method for detecting bTB infections, its effectiveness needs to be assessed regularly [19]. Failure to detect infected animals at slaughter will potentially enable the further transmission of bTB infection from and within the herds the slaughtered animals originate from, especially in endemic areas with a low frequency of skin testing as is the case in parts of Great Britain [27]. Several studies [16, 19, 22, 26] have been conducted to measure the variation between factories in the proportion of slaughtered animals from which a sample was submitted to the laboratory (‘submission risk’) and the variation in the proportion of those samples that were subsequently confirmed to be due to M. bovis (‘confirmation risk). For the period 2003–2004, submission and confirmation risks in Ireland were at average 0.22% and 64.4% [16], for the period 2005–2007 these were 0.25% and 68.5% [26]. The submission risk varied from 0.08% to 0.58% [16] and from 0.03% to 0.53% [26] between factories that have slaughtered at least 10,000 animals. This variation could be due to confounding factors, i.e., age, sex, the herd of origin, the season of slaughter, bTB history of the herd, and the geographical risk that were taken into account in the estimation of the factory’s relative probability to submit a sample. Risk factors associated with FL detection might differ over time due to changes in the composition of the national cattle population [28] and epidemiological situation [5].

Therefore, the objective of this study was to reassess the relative effectiveness of Irish factories in submitting FL and the confirmation rate of those FLs from slaughtered cattle over the period 2014–2018. Also, risk factors at the animal, herd, and factory level for the submission and confirmation of FL were identified and quantified. The results from this study provide an evidence base for the evaluation of the effectiveness of this component of bTB surveillance in Ireland. Next to that, it will inform policy development and operational oversight to improve the detection of infected herds, support the updating of the bTB eradication programme and contribute to the reduction of bTB levels in Irish cattle.

Materials and methods

Dataset

The data for this study were obtained from the Animal Health Computer System (AHCS) of Department of Agriculture, Food and Marine (DAFM), Ireland. The original dataset includes information on 6,611,854 animals slaughtered in 35 Irish factories from 2014 to 2018. The dataset contains records of animal data, tuberculin tests, and laboratory results. The recorded animal data includes animal identification numbers, age, sex, breed, date of birth (DOB), date of death (DOD), and whether a lesion was detected and submitted for laboratory confirmation. Data on the herd history of annual tuberculin testing based on the annual SICTT concerned the number of animals tested per herd, last testing date, herd size, herd type, herd identification number, and herd location based on county. The carcasses of slaughtered animals are routinely examined by veterinary inspectors for presence of lesions, as part of standard food safety post-mortem veterinary examinations. The inspection included visual examination, palpation, and incision of several organs (e.g., lungs, heart, and liver) and lymph nodes such as parotid, submandibular, retropharyngeal, apical, bronchial, tracheobronchial, mediastinal, hepatic, and mesenteric lymph nodes. When a suspected lesions was observed, tissue samples were collected and sent to the laboratory, where samples were processed for histopathological staining and/or bacterial culture to confirm bTB infection.

Data analysis

All collected data for this study were double-checked for redundancy, missing values and then assembled for analysis. Due to many animal breeds, breed was categorized as cross breed and pure breed. Animal movement was categorized into purchased and homebred animals based on the herd identification number. If the animal was born and sent to slaughter from the birth herd, it was categorized as “homebred”, if not as “purchased”. All potential risk factors included in the statistical analysis were categorized as shown in Table 1.

Table 1

Potential risk factors considered to have associations with FL submission and confirmation from animals slaughtered in Irish factories from 2014–2018.

Risk factors	Categories
Factory	Per factory
Year	Per year slaughter (2014–2018)
Month of slaughter	Per month (January–December)
Animal-related factors
Age (year)	Per year category (1 to ≥ 8 years)
Sex	Male; Female
Animal movement	Homebred; Purchased
Breed	Cross breed; Pure breed
Ever inconclusive	Yes; No
Ever severe inconclusive	Yes; No
Herd-related factors
Herd type	Beef; Dairy; Others; Suckler;
Herd size	Log transformed
BTB history (past 3 years)	Yes; No

The analytical method was in accordance with previous studies [16, 26]. The relative performance of each factory in detecting bTB lesions was based on the submission of suspected lesions and the laboratory confirmation of submitted lesions as bTB positive. In the first step, descriptive analysis was carried out to summarize the risk of FL submission and confirmation per factory, and the Spearman rank correlation between submission and confirmation risk was estimated. In addition, the frequency distribution and proportion of submitted and confirmed FL were calculated for each category of the risk factors. All data editing and descriptive analyses were performed in SAS on Demand (SAS Institute Inc). The effect of each risk factor on the FL submission and confirmation risk was analysed with univariable logistic regression (with herd as random effect), followed by multivariable mixed effects logistic regression using Stata SE 16.1 (Stata Corp, USA), see Eq 1. Where π is the probability of having a lesion submitted (or confirmed if submitted) if an animal of herd j is slaughtered in factory i; β0, β, β are the regression coefficients for the intercept, the factories (F) and all the covariates (C) respectively and where μ represent the random effect of herd j. In the univariable model the covariates were not included.

The crude and adjusted odds ratio (OR) of each factory, as obtained from univariable and multivariable analysis, were ranked accordingly from the highest to lowest OR, and a comparison of crude and adjusted ranking were made to determine whether the adjustment with potential confounding risk factors and herd effect affected the OR estimates and the ranking of factories. The comparison of crude and adjusted factory ranking and OR were visualized in forest plots using Microsoft Excel (Microsoft Corporation, USA).

Results

Descriptive results

The dataset of routine slaughterhouse surveillance consisted of 6,611,854 animals slaughtered in 35 Irish factories from 2014 to 2018 (S1 Table). Records of 33,609 (0.51%) animals were not used in the analysis: 6202 animals due to four small factories (slaughter <10,000 animals during the study period), 2716 animals due to missing values for at least one variable (i.e., age, number of skin tests, breed, herd type, herd size), and 24,901 animals were below one-year-old (some of these animals had combined reasons of exclusion with small factory and missing values). Twenty-five (0.07%) of excluded animals had FL submitted, and 16 (64%) of those FL were confirmed as bTB positive. The final dataset for statistical analysis consisted of 6,578,245 animals originating from 84,429 herds that were slaughtered in 31 Irish factories. Many herds sent their animals to more than one factory during the study period (up to 17 factories). A total of 13,337 (0.20%) animals were suspected of having FL, of which 5431 (40.72%) were confirmed as bTB positive in the laboratory. The correlation between crude submission and confirmation risk of the factories were not statistically significant (Spearman ρ = -0.074, P = 0.69). The crude submission risk per factory ranged from 0.01 to 0.43%, average 0.20% and the confirmation risk of submitted lesions as bTB positive ranged from 0.00 to 61.84%, average 40.72% (Table 2). The submission and confirmation risk of factories also varied over years (Figs 1 and 2). Overall, the proportion of bTB confirmed animals based on FL submission was 0.08% in the 2014–2018 period.

Table 2

Distribution of animals slaughtered, submission risk, and confirmation risk of submitted lesions over factories included in the analysis.

Factory	Total slaughtered	Number submission	Submission risk (%)	Number confirmation	Confirmation risk (%)
S01	222,696	952	0.43	348	36.55
S02	73,089	268	0.37	130	48.51
S03	310,734	1031	0.33	492	47.72
S04	210,407	627	0.30	273	43.54
S05	258,947	735	0.28	239	32.52
S06	232,000	621	0.27	148	23.83
S07	28,844	76	0.26	24	31.58
S08	296,572	752	0.25	300	39.89
S09	319,900	801	0.25	366	45.69
S10	316,175	756	0.24	392	51.85
S11	386,745	888	0.23	301	33.90
S12	238,572	542	0.23	257	47.42
S13	224,124	495	0.22	189	38.18
S14	233,147	479	0.21	244	50.94
S15	225,286	433	0.19	103	23.79
S16	69,870	132	0.19	48	36.36
S17	229,438	428	0.19	115	26.87
S18	250,963	416	0.17	146	35.10
S19	16,907	28	0.17	9	32.14
S20	333,640	489	0.15	154	31.49
S21	229,964	310	0.13	162	52.26
S22	307,640	412	0.13	240	58.25
S23	106,646	141	0.13	61	43.26
S24	51,524	65	0.13	19	29.23
S25	281,415	350	0.12	137	39.14
S26	262,645	308	0.12	131	42.53
S27	145,368	152	0.10	94	61.84
S28	223,734	209	0.09	76	36.36
S29	366,590	338	0.09	175	51.78
S30	111,237	101	0.09	58	57.43
S31	13,426	2	0.01	0	0
Total	6,578,245	13,337	0.20	5431	40.72

Fig 1

Submission risk (percentage) of lesions for 31 Irish factories per year (2014–2018).

Fig 2

Confirmation risk (proportion) of lesions submitted by 31 Irish factories per year (2014–2018).

The majority of animals slaughtered in Irish factories were two years old (48.50%), and 8.34% were equal to or more than eight years old. The submission risk increased as the animal gets older, ranging from 0.15 to 0.42%, while the confirmation risk ranged from 28.31 to 60.55%. Female animals had a higher submission and confirmation risk than male animals (0.25 vs. 0.17% and 48.05 vs. 31.02%, respectively). The average age at slaughter of females was 54.25 months and 27.03 months for males. Purchased animals had slightly higher submission and confirmation risk than homebred animals (0.21 vs. 0.20% and 41.33 vs. 39.21%, respectively). A total of 11,526 (0.18%) of slaughtered animals had at least one inconclusive result in routine tuberculin testing, and the submission risk of these animals was strongly increased (14.34 vs. 0.18%), although the confirmation risk was lower compared to animals that never had an inconclusive test result (32.49 vs. 41.89%). 17,877 (0.27%) of the slaughtered animals had been tested at least once as severe inconclusive result, and their submission and confirmation risk increased compared to animals that never had a severe inconclusive test (1.40 vs. 0.20% and 85.60 vs. 39.86%, respectively). Submission and confirmation risk were higher in animals originating from herds with bTB history in the past three years compared to animals originating from herds without bTB history in the past three years (0.41 vs. 0.12% and 69.73 vs. 2.44%, respectively). The descriptive results for all potential risk factors are presented in Table 3.

Table 3

Univariable and multivariable (adjusted for covariates and random herd effect) logistic regression results of submitted (n = 6,578,245) and confirmed (n = 13,337) bTB lesions from animals slaughtered in Irish factories in the period 2014–2018.

Variable	Number slaughter	Submission (%)	Confirmation (%)	Univariable		Multivariable
				Submission OR; 95% CI	Confirmation OR; 95% CI	Submission OR; 95% CI	Confirmation OR; 95% CI
Year
2014	1,354,005	0.20	46.14	Ref.	Ref.	Ref.	Ref.
2015	1,273,270	0.21	42.40	1.08; 1.02–1.14	0.79; 0.67–0.91	1.13; 1.07–1.20	0.81; 0.67–0.98
2016	1,323,348	0.21	37.85	1.08; 1.02–1.14	0.62; 0.53–0.72	1.16; 1.10–1.23	0.73; 0.60–0.88
2017	1,399,721	0.20	39.38	1.05; 1.00–1.11	0.70; 0.60–0.81	1.13; 1.07–1.19	0.82; 0.67–0.99
2018	1,227,901	0.20	37.87	1.05; 1.00–1.11	0.65; 0.56–0.76	1.13; 1.06–1.20	0.80; 0.66–0.99
Age (year)
1	1,845,788	0.15	28.13	Ref.	Ref.	Ref.	Ref.
2	3,190,707	0.16	37.07	1.11; 1.06–1.17	1.68; 1.46–1.94	1.11; 1.05–1.17	1.75; 1.46–2.09
3	363,954	0.28	36.37	1.91; 1.77–2.06	1.64; 1.33–2.03	1.73; 1.59–1.87	1.38; 1.05–1.80
4	176,970	0.31	42.86	2.20; 2.00–2.42	2.39; 1.84–3.11	1.84; 1.66–2.04	2.3; 1.62–3.26
5	160,529	0.31	41.18	2.18; 1.97–2.40	2.22; 1.69–2.91	1.70; 1.53–1.90	1.71; 1.20–2.44
6	150,557	0.35	48.66	2.47; 2.25–2.72	3.22; 2.47–4.20	1.88; 1.69–2.09	3.06; 2.12–4.40
7	140,981	0.36	54.69	2.52; 2.29–2.78	4.54; 3.46–5.97	1.86; 1.67–2.08	3.42; 2.38–4.93
≥8	548,759	0.42	60.55	2.91; 2.75–3.09	6.28; 5.25–7.52	1.91; 1.77–2.06	4.05; 3.17–5.18
Sex
Female	3,017,843	0.25	48.50	Ref.	Ref.	Ref.	Ref.
Male	3,560,402	0.17	31.03	0.67; 0.64–0.70	0.36; 0.32–0.40	1.05; 1.00–1.10	0.66; 0.56–0.76
Animal movement
Homebred	1,777,728	0.21	39.21	Ref.	Ref.	Ref.	Ref.
Purchased	4,800,517	0.20	41.33	0.97; 0.93–1.01	1.15; 1.02–1.28	0.96; 0.91–1.01	0.96; 0.81–1.14
Breed
Cross breed	4,716,660	0.20	39.82	Ref.	Ref.	Ref.	Ref.
Pure breed	1,861,585	0.21	42.96	1.03; 0.99–1.08	1.22; 1.10–1.37	0.80; 0.76–0.83	1.04; 0.89–1.22
Ever inconclusive
No	6,566,719	0.18	41.89	Ref.	Ref.	Ref.	Ref.
Yes	11,526	14.34	32.49	99.04; 93.19–105.26	0.64; 0.55–0.75	74.24; 69.39–79.43	0.44; 0.36–0.54
Ever severe inconclusive
No	6,560,368	0.20	39.86	Ref.	Ref.	Ref.	Ref.
Yes	17,877	2.17	85.60	6.89; 6.02–7.81	17.94; 11.27–28.54	2.83; 2.44–3.27	4.48; 2.66–7.54
Herd type
Beef	2,435,812	0.20	40.74	Ref.	Ref.	Ref.	Ref.
Dairy	1,267,432	0.20	37.91	1.06; 1.00–1.12	0.88; 0.75–1.02	0.83; 0.77–0.90	1.08; 0.85–1.38
Other	520,788	0.18	33.19	0.85; 0.78–0.93	0.61; 0.49–0.77	0.99; 0.90–1.08	1.11; 0.81–1.51
Suckler	2,354,213	0.21	43.51	1.13; 1.07–1.18	1.21; 1.07–1.38	1.08; 1.02–1.14	1.31; 1.11–1.55
Herd size				0.98; 0.96–0.99	0.96; 0.59–0.81	0.95; 0.93–0.96	0.82; 0.78–0.86
BTB history (past 3 years)
No	4,722,722	0.12	2.44	Ref.	Ref.	Ref.	Ref.
Yes	1,855,523	0.41	69.73	4.87; 4.66–5.09	355.80; 2.65–478.26	4.46; 4.28–4.66	319.90; 237.98–430.04
Months of slaughter
January	562,269	0.19	37.55	Ref.	Ref.	Ref.	Ref.
February	430,767	0.18	42.56	0.93; 0.85–1.02	1.31; 1.01–1.71	0.94; 0.86–1.04	1.10; 0.79–1.53
March	562,055	0.18	39.36	0.97; 0.89–1.06	1.09; 0.85–1.38	0.98; 0.90–1.07	1.13; 0.83–1.53
April	527,150	0.20	38.06	1.07; 0.98–1.16	1.06; 0.83–1.34	1.04; 0.95–1.14	1.01; 0.74–1.36
May	534,771	0.19	38.73	1.02; 0.93–1.11	1.15; 0.90–1.46	0.94; 0.86–1.03	1.02; 0.75–1.39
June	536,051	0.21	40.37	1.11; 1.02–1.21	1.07; 0.84–1.36	1.01; 0.93–1.10	0.84; 0.62–1.13
July	558,296	0.22	40.87	1.16; 1.07–1.26	1.24; 0.98–1.56	1.05; 0.96–1.15	1.11; 0.83–1.50
August	588,359	0.21	39.34	1.09; 1.01–1.19	1.12; 0.89–1.42	1.02; 0.94–1.11	1.01; 0.75–1.35
September	622,015	0.21	41.42	1.10; 1.01–1.19	1.23; 0.98–1.55	1.05; 0.97–1.15	1.14; 0.85–1.53
October	617,809	0.20	42.92	1.05; 0.97–1.14	1.29; 1.02–1.63	1.01; 0.93–1.10	1.39; 1.03–1.87
November	626,507	0.20	42.93	1.04; 0.95–1.13	1.38; 1.10–1.75	0.99; 0.91–1.08	1.34; 1.00–1.81
December	412,196	0.22	44.86	1.14; 1.04–1.25	1.48; 1.16–1.91	1.11; 1.02–1.22	1.55; 1.13–2.13
Total	6,578,245	0.20	40.72

Modeling outcomes

Submission

The submission OR of factories in the univariable and multivariable analysis (including all covariates) with herd as random effect are shown in Fig 3. The relative factory ranking did not change greatly: only 6 factories swapped between quartiles of the ranking. The maximum change in factory ranking was 6 positions and the average being 2 positions (S2 Table). Herd as random effect explained 15% of the residual variance (Intraclass correlation (ICC) = 0.15; 95% CI 0.14–0.16) in the multivariable model. The odds of FL submission increased with the increase of animal’s age at slaughter, and the OR of animals over eight years old was 1.91 compared to animals of 1–2 years old. Males had lower odds of having FL submitted than females (OR 0.67, 95% CI 0.64–0.70) in the univariable analysis, but the odds increased to 1.05 in the multivariable analysis. Ever inconclusive tested animals under standard interpretation had 74.24 (95% CI 69.39–79.43) times higher odds of FL submission compared to never inconclusive animals in the multivariable analysis. The odds were much lower for animals that were defined as inconclusive based on the severe interpretation of the tuberculin skin test (OR: 2.83; 95% CI 2.44–3.27). Pure breed animals had lower odds of FL submission (OR 0.80; 95% CI 0.76–0.83) than cross breed animals. Animals originating from suckler herds had slightly higher odds of having submitted FL than animals from beef herds (OR 1.08, 95% CI 1.02–1.14), while dairy herds had lower odds of having FL (OR: 0.83; 95% CI 0.77–0.90). Herd size was negatively associated with FL submission; an increase in log herd size by one unit (mean 4.3, median 4.6, maximum 7.6) reduced the odds with a factor 0.95 (95% CI 0.93–0.96). A history of bTB breakdown in the past three years increased the odds of having a submitted FL (OR 4.46; 95% CI 4.28–4.66). The submission OR for all potential risk factors are presented in Table 3.

Fig 3

Comparison of factory ranking based on OR derived from univariable and multivariable (adjusted for covariates and random herd effect) logistic regression results of bTB suspected lesion submission from animals slaughtered (n = 6,578,245) in Irish factories in the period 2014–2018.

S09 = reference category.

Confirmation

The confirmation ORs of factories of the univariable and multivariable model (including all covariates) are shown in Fig 4. The relative ranking of some factories substantially changed after adjustments for covariates: nine factories swapped between quartiles of the ranking and the change in factory ranking was up to 26 positions, the average being 4 positions (S2 Table). 30% of the residual variance of the multivariable model was explained by the herd effect (Intraclass correlation (ICC) = 0.30; 95% CI 0.26–0.35). Submitted FL from older animals had higher odds of being confirmed as bTB positive in the laboratory. The odds of confirming FL in animals over eight years old were 4.05 times (95% CI 3.17–5.18) higher than the odds of animals 1–2 years old. Males had lower odds of having FL confirmed than females (OR 0.66; 95% CI 0.56–0.76). The confirmation odds were also lower for FL originating from animals with inconclusive results of tuberculin skin tests (OR 0.44; 95% CI 0.36–0.54), although the odds of confirming FL from severe inconclusive animals was increased with a factor 4.48 (95% CI 2.66–7.54). The confirmation odds were slightly higher for FL originated from cross breed animals (OR 1.04; 95% CI 0.89–1.22) than pure breed, although it was not statistically significant. FL from animals of suckler and dairy herds had 1.31 (95% CI 1.11–1.55) and 1.08 (95% CI 0.85–1.38) higher odds to be confirmed as bTB positive than FL from animals of beef herds. Increasing log herd size by one unit (mean 4.3, median 4.6, maximum 7.6) lowered the odds of FL confirmation by a factor 0.82 (95% CI 0.78–0.86). Compared to the herds with no bTB breakdown in the past three years, the odds of confirming an FL were 319 times (95% CI 237.98–430.04) higher in animals from herds with recent bTB history. The confirmation OR for all potential risk factors are presented in Table 3.

Fig 4

Comparison of factory ranking based on OR derived from univariable and multivariable (adjusted for covariates and random herd effect) logistic regression results of bTB confirmation of lesions submitted from animals slaughtered (n = 13,337) in Irish factories in the period 2014–2018.

S09 = reference category.

Discussion

The main aim of this study was to quantify differences in the relative effectiveness of Irish factories in detecting suspected bTB lesions among the animals slaughtered from 2014 to 2018. This study also assessed the association between potential risk factors and the submission and confirmation risk of submitted lesions.

The submission risk of FL in 31 Irish factories ranged from 0.01 to 0.43%, which means that out of 10,000 slaughtered animals, 1 to 43 animals that had a suspected FL were submitted to the laboratory for bTB confirmation, the average being 20 animals per 10,000 slaughtered animals. Similar results were obtained in a previous study over the period 2003–2004 from which 0 to 58 animals with an average of 22 animals per 10,000 slaughtered animals was reported [16]. In the period 2005 to 2007, these numbers were 0 to 52 animals with an average of 25 animals per 10,000 slaughtered animals [26]. A nine-fold, seven-fold, and five-fold increase after exclusion of factories that submitted less than 10 animals were reported from earlier studies [16, 26, 29]. This is in line with the current study as the difference is around 5-fold (0.09 vs. 0.42%) if the same criteria are used (Table 2). Furthermore, the confirmation risk of submitted FL ranged from 0.00 to 61.84% with an average of 40.72%, which is lower than the average confirmation risk reported in those previous studies of 67.2% [30], 63% [16], and 67.5% [26]. The relatively lower confirmation risk in the present study is consistent with lower bTB levels in Ireland during the study period compared to those previous study periods.

The FL submission and confirmation risk varied between factories in the univariable and multivariable analysis, and might show unequal practices in lesions detection between factories [26]. The crude and adjusted submission ranking of factories were not much different, which indicates that herd and animal factors did not contribute substantially to the variation of submission risk. However, a substantial difference in crude and adjusted confirmation ranking of some factories may indicate that animal and herd-related factors contribute substantially to the variation in FL confirmation between factories. From previous studies [16, 26, 30] it was concluded that the variation of submission and confirmation risk was not related to animal and herd-related factors, but to assigned factory-related factors that may influence the inspection performances in the slaughterhouse. Such factors include line speed, lighting, equipment, and competency of the veterinary inspectors [31, 32]. Competencies and skills of the veterinary inspectors in each factory in recognizing potential bTB lesions are important during the meat inspection. Suboptimal performance by inspectors might lead to missed lesions and the herds with truly infected animals remaining undetected. This especially may occur if the infection is in the initial stage where granulomatous lesions are frequently absent or relatively small, making it hard to detect lesions by the naked eye [33].

Several animal and herd-related factors have shown to influence bTB infection [16,19, 20, 26, 34], we are cognisant that these factors (e.g., herd type, breed, sex, and age) are somewhat intertwined. In this study, as expected, FL are more likely to be detected and confirmed in older animals, which in agreement with the findings of [16, 26, 30] in the Irish setting. Older animals will have spent a longer time in the herd, which prolonged exposure to infected animals and contaminated environment [16, 34, 35]. In addition, because bTB is a chronic disease, a longer lifespan gives more time to develop visible granulomatous lesions [30]. An inverse relationship of FL submission and confirmation risk was present in male and female animals, and this is potentially confounded by age. FL were more likely to be disclosed in males; however, the confirmation odds were significantly lower. Females have been reported to have higher odds of being bTB positive, whether detected by routine skin testing [36] or at slaughter [16]. Different submission and confirmation risks between sex are possibly due to different management practices; for instance, dairy cows usually have a longer lifespan due to milk production [37] and therefore may have more developed lesions which are more easily detected or confirmed at slaughter.

Animals that ever had at least once inconclusive or severe inconclusive tuberculin skin result had higher odds to have a submitted FL. FL from ever inconclusive animals had lower odds to be confirmed as bTB positive (OR = 0.44); in contrast, the confirmation odds was higher for FL from severe inconclusive animals (OR = 4.48) (Table 3). A possible explanation is the policy in handling inconclusive animals. In Ireland, during the study period, animals that had an inconclusive skin result under standard interpretation must be either re-tested 42 days later or slaughtered. If farmers choose the option to have them slaughtered, lymph nodes are submitted to the laboratory regardless of the presence of visible lesions, which explains the increased submission risk and the lower confirmation risk (pers. comm. Philip Breslin, DAFM).

In the models we fitted a breed variable representing whether animals were reported within the AHCS dataset as a “pure breed” or “cross breed”. Ideally, the actual breed or breed functional class (e.g. dairy breed, beef breed, dual purpose) could have been used, but exploratory analysis suggested that there was strong corelations between these classifications and “herd type” variable. Therefore, to avoid variation inflation, we instead dichotomised breed based on their reported genetics mix. It should be noted that this variable may have some limitations given the data on breed heritage which is self-reported by the farmer within the AHCS dataset. Furthermore, it is not fully known whether breed crosses are more or less susceptible to M. bovis infection, or that they differ in disease progression. However, some evidence suggests differences in breed heterozygosity were associated with lower susceptibility to bTB in Ireland [38], and in prevalence between crosses and local breeds in Ethiopia [39] and India [40], for example.

The odds of FL submission and confirmation were negatively associated with herd size. A similar conclusion was made in previous research undertaken at animal-level [35, 41, 42]. This is in contrast with the well-known increased risk of herd-level bTB breakdown in larger herds [37, 43]. Higher odds of having FL in smaller herds is suggested to be related to management practices in Ireland, where animals in small herds usually have close contact with other animals on the farm [41], while the larger herds are being split in cohorts and reduce the contact with infected animals [35]. Animals from suckler and dairy herds were more likely to have FL confirmed as bTB positive than beef herds. This can be due to the genetics or cattle breed [44], longer contact of suckler calves and their dam [35], or due to the longer lifespan of dairy cattle [37]. The history of bTB infection was significantly related to the submission and confirmation risks of FL. Animals from herds with a history of bTB in the past three years were more likely to have a submitted and confirmed FL. A comparable finding was obtained by [16]; however, that study considered the history of OTF years where animals from herds with 0–3 years OTF history have higher odds of having lesions confirmed than animals from herds >4 years OTF history. Herds with a history of bTB also had a higher risk for restriction in the future [45]. In the herds with bTB history, it may be possible that infected animals remain undetected with the routine skin tests and transmit the infection to other animals before they are slaughtered. It might also be due to a re-introduction of the infection by the purchase of latently infected cattle or by a wildlife reservoir.

Conclusion

In comparison to the similar studies in Ireland during 2003–2005 [16] and 2005–2007 [26], the FL submission remained steady, but the confirmation risk was lower, consistent with a reduction in overall bTB levels in Ireland across the study periods. The relative submission ranking of factories did not change much after adjustment for animal and herd-related factors, suggesting FL submission variation between factories is mainly due to factory-related factors. The crude and adjusted ranking of some factories was substantially different for FL confirmation, which indicates that animal and herd-related factors may contribute to the variation of FL confirmation risk between factories.

(ZIP)

Click here for additional data file.

28 Jun 2022

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: https://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols.

We look forward to receiving your revised manuscript.

Kind regards,

Rebecca Lee Smith, D.V.M., M.S., Ph.D.

Academic Editor

PLOS ONE

Journal Requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf  and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

2. Thank you for stating the following in the Acknowledgments Section of your manuscript:

“Further thank goes to the Indonesia Endowment Fund for Education (LPDP) that granted a scholarship for RRMH to study at Wageningen University & Research”

We note that you have provided additional information within the Acknowledgements Section that is not currently declared in your Funding Statement. Please note that funding information should not appear in the Acknowledgments section or other areas of your manuscript. We will only publish funding information present in the Funding Statement section of the online submission form.

Please remove any funding-related text from the manuscript and let us know how you would like to update your Funding Statement. Currently, your Funding Statement reads as follows:

“This study was funded by Indonesia Endowment Fund for Education (LPDP, https://lpdp.kemenkeu.go.id), Ministry of Finance, the Republic of Indonesia, as part of the scholarship awarded to RRMH. The funder had no role in study design, data collection, and analysis, decision to publish, or preparation of the manuscript.”

Please include your amended statements within your cover letter; we will change the online submission form on your behalf.

3. In your Data Availability statement, you have not specified where the minimal data set underlying the results described in your manuscript can be found. PLOS defines a study's minimal data set as the underlying data used to reach the conclusions drawn in the manuscript and any additional data required to replicate the reported study findings in their entirety. All PLOS journals require that the minimal data set be made fully available. For more information about our data policy, please see http://journals.plos.org/plosone/s/data-availability.

Upon re-submitting your revised manuscript, please upload your study’s minimal underlying data set as either Supporting Information files or to a stable, public repository and include the relevant URLs, DOIs, or accession numbers within your revised cover letter. For a list of acceptable repositories, please see http://journals.plos.org/plosone/s/data-availability#loc-recommended-repositories. Any potentially identifying patient information must be fully anonymized.

Important: If there are ethical or legal restrictions to sharing your data publicly, please explain these restrictions in detail. Please see our guidelines for more information on what we consider unacceptable restrictions to publicly sharing data: http://journals.plos.org/plosone/s/data-availability#loc-unacceptable-data-access-restrictions. Note that it is not acceptable for the authors to be the sole named individuals responsible for ensuring data access.

We will update your Data Availability statement to reflect the information you provide in your cover letter.

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: No

Reviewer #2: Yes

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: No

Reviewer #2: Yes

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: No

Reviewer #2: Yes

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: Main considerations

The submitted manuscript reports a study on the submission rate of potentially bovine TB infected lesions from cattle slaughterhouses in Ireland, and on the subsequential bTB confirmation rates of these samples. The study aims to compare the submission and confirmation rates (they are called “risks” in the manuscript, see last point of the main considerations) across different slaughterhouses, and evaluate the risk factor which can have an influence on these rates. In particular, they compare the odds ratio of each slaughterhouse in an univariable model and with a multivariable model, including the number potential factors that might play a role in determining the slaughterhouses confirmation rates. While the subject is of interest there are some pitfalls which undermine the robustness and the depth of these findings.

First, I think there is a confusion between the concepts of “risk” and “rate”. As it stands, I found it very odd to consider that the chance of a sample to be submitted to the lab as a “risk” (same for “confirmation”). I’d suggest to change these to “submission rate” and “confirmation rate”.

A crucial but neglected aspect is the sampled population spatial distribution, i.e. where the slaughterhouses get the cattle from. Since bTB is not homogeneously spread in Ireland, the location where becomes very important. This is exacerbated by the wildlife interface spread, which makes some areas widely different to others in term of potential bTB prevalence. By neglecting so, it corresponds to assume that all the cattle are homogeneously mixed in Ireland, and that bTB is homogeneous as well, or, that all the slaughterhouses get the cattle form the same exact source. Including cattle geographical factors (e.g. farm coordinates, local TB rates) would be important for the validity of this study, otherwise comparing the slaughtrhouses submission and confirmation rates does not provide much information.

Similarly, no analysis was performed ex-post about what might cause different rates of submission and confirmation in the slaughterhouses. Do ones with similar rates cluster in space, for the type of source herds or else? It would be also interesting comparing the location of the slaughterhouses in relation to the local bTB prevalence, but I understand that location information might be subject to privacy restrictions. Nothing prevent them to be analysed, even if they are not shown on a map.

Starting by saying that I am not a statistician, I don’t understand the logic behind the model usage here. In my understanding univariable models are run with a single explanatory variable at a time, and then the most significant ones are run in a multivariable model. At first my best guess was that the univariable model included only the slaughterhouse (as a fixed effect), and that the multivariable model included all the variable in table 1 as fixed effects, and the slaughterhouses as a random effect. However, I realised that the “univariable” model included many variables indeed (see Table S4 and S5). Was the univariable model just a multivariable model but run independently for each slaughterhouse? Or was the slaughtrhouses factor just neglected in the "univariable" model? If this is the case, was the comparison between “univariable” and “multivariable” just a methodological aspect?

The difficulty in understanding what was done is also caused by the absence of any model formula. Even if these are well known methods, reporting a rigorous mathematical formula would help the readers (and the reviewers) in understanding what was done. Finally, no model metric was reported (AIC, BIC). Were all models performing in a similar way? How well these models explained the observations?

Finally, I suggest a thorough read of the manuscript before potential resubmissions.

Minor comments

General: the term “factory lesions” sounds odd, it seems like the lesions happened at the factory. I can live with that, but maybe “internal lesions” or something else could be more straightforward for the readers.

Abstract: it would be more beneficial for the readers to have some more background and broader conclusions and considerations in this section, rather than delving into the details of the results.

Line 49: in Europe only?

Line 56: remove “thus”.

Line 58: I know that the presence of bTB deer is still controversial, but it is worth mentioning (see Crispell et al., 2020).

Line 65: the badger culling program?

Line 81: why an interval is reported here? Is it the percentage of breakdowns per year?

Lines 80-83: would it be possible to report all three data in the same format, like all number or percentages (better) of breakdowns disclosed? Otherwise they are difficult to compare.

Line 92: remove “the” in front of Ireland.

Lines 93-94: confusing, please just report the range.

Lines 99-105: these sentences are convoluted and hard to read.

Line 113: is DOD because of “date of death”?

Line 127: slaughtered in the same herd or sent to slaughter from the birth herd?

Line 137: which distribution was calculated? And where is it reported?

Table 1: are slaughterhouses Z, ZA, ZB, ZC, ZD, ZE related or there are just not enough letters? Did you consider to switch to number or a code instead (S01, S01 or any)?

Table 1 and Figure 1: are the slaughterhouses in random order? Would it be more valuable for the readers if they were ranked according to number of slaughtered animals (or some other metric), instead?

Table 1: how solid is the inclusion of slaughterhouse T in the study, given only 2 samples submitted to the lab?

Lines 173-174: confusing, please rephrase.

Line 180: “these animals” are the ones with at least one inconclusive test?

Line 185: were “higher”?

Line 186: how is the 3-year threshold being selected? Does it make a difference if the 2 or 4 years would be selected to defined the past bTB history? This sensitivity analysis would be material for supplementary information.

Line 224: “lower”?

Lines 253-261: I struggle to understand the point of this paragraph, in particular 257-261. Why the VOI procedure should make any difference, if it’s not in place in Ireland?

Lines 271-275: were these factors addressed in this analysis?

Lines 284-287: herd type, breed, sex and age are intimately intertwined factors in modern cattle industry. Since they are mostly treated as stand-alone factors, I would suggest to integrate them at least in the discussion, so to provide the readers a clearer picture.

Lines 323-332: this paragraph is a repetition from the introduction, rather than discussion material.

S1: I might have missed this in the text, but why some factories have been excluded from the study?

Reviewer #2: The manuscript “revisiting the relative effectiveness of slaughterhouses in Ireland to detect tuberculosis lesions in cattle (2014-2018)” updates the analyses carried out in the past regarding the differences between slaughterhouses in Ireland in the rate of submission and laboratory confirmation of tuberculosis-compatible lesions. Given the importance of passive surveillance in the eradication programs for bovine tuberculosis in countries where the disease is at (more or less) reduced levels, the continued evaluation of its performance is an objective worth pursuing, and therefore the topic under study here has scientific merit. The analytical framework used is well established and was used in the past with the same objective, and the interpretation of the results is sound. Still, the addition of some details in terms of methodology and results could help to better understand what was performed, and what the results were.

Comments:

- Line 21-22: the term “ever inconclusive or severe inconclusive skin test reactor” is not very straightforward, suggest replacing it by “ever reactor in the standard or severe skin test” or similar (more similar to what was used in lines 30-31)

- Line 46-47: I would suggest the authors to refer to bovine tuberculosis as the disease caused by members of the M. tuberculosis complex in bovine rather than the disease caused by M. bovis in any host species (though both definitions are widely used in the literature, I think we would not refer to bovine brucellosis as the infection by B. abortus in sheep, and the former definition is more aligned with current regulations in the EU).

- Line 51: suggest replacing “in” by “by” (since countries did not gain the OTF status precisely in 2019)

- Lines 94-97: Given that in the papers mentioned in the previous lines (refs 15 and 20) at least some of the covariables mentioned here (age, gender…) were considered in the multivariable models, I would expect that variations in their distributions depending on the slaughterhouse should not explain the difference in submission risk per slaughterhouse (that is the point of considering them in the analysis, isn’t it?).

- Line 113: I think gender is a trait only applicable to humans, and sex should be used here (as e.g. used in Table 1).

- Line 114: for purchased animals: did the herd history include all the annual tuberculin testing in all herds where the animal had stayed or only the last one prior to be shipped to the abattoir? (the last testing date is mentioned in line 115 but I wonder if this refers to the whole information on skin testing).

- Data analysis: Herd size is an interesting covariable but I wonder if it makes sense to assume a linear relationship between the (logit of) the probability of submission/confirmation and the number of animals in the herd. Did the authors consider categorizing in some way this variable and see whether it improved (or not) model fit? Although categorizing a continuous variables has its own drawbacks, in terms of interpretation it may be easier to see increase in the risk in “large” vs. “small” herds (just a suggestion). Also, it seems (based on Table S4) it was log-transformed; if so, this should be indicated in the methodology section.

- Data analysis (II): I would be interested in knowing the view of the authors regarding considering factory as a random rather than as a fixed effect. Though I see the point in considering it a fixed effect, given the very large number of categories and the potential objective of characterizing variability across slaughterhouses assuming the belong to some sort of population, coupled with the view of the database as a hierarchical structure (with animals clustered in abattoirs). Note that I don’t think there is anything wrong with the approach used and I am not requesting the authors to change it, but I would like to know if this was considered at some point or if the authors have a strong view against this option.

- Line 155: I was surprised to see animals <1 year were excluded but looking at S2 I realize they represent a very small fraction of the population. In other countries it may be not so unusual to slaughter animals below one year (10-11 months-old), so perhaps the fact that animals below 1 year were excluded could be added in the material and methods as an exclusion criterion

- Line 153: suggest to specify that there are four small factories (I think there are four based on line 158, 31 factories after exclusion of small ones).

- Lines 161-162: what does this correlation refer to? The crude submission and confirmation risks per factory? This could be perhaps added as a supplementary graph (and although I assume it is considered included in the “descriptive analysis” part on the material and methods, also be specified there as an analysis performed).

- Line 175: suggest replacing “ranged” by “ranging”

- Modeling outcomes: I would suggest to start both submission and confirmation sections by indicating which variables (all considered, I pressume?) were included in the final model. Also, I would strongly suggest to add a table in the main text with the adjusted OR coefficients (along with some of the descriptives included in Table S2?) for these covariables (something similar to Table 1 in the paper by Frankena et al). Though this is included in table S4, I think it is very useful information that readers would appreciate having in the main text (and could be easily integrated in Table 1, that currently conveys only limited information).

- Line 191-192 and 216-217: given that factory-specific ORs are calculated based on the difference with the baseline abattoir, I am not sure how to interpret these sentences (differences between factories become larger once covariates were considered? But this would be only considering the reference category, wouldn’t it? Perhaps other differences decrease?)

- Line 251: suggest replacing “difference” by “increase” (since I think this is what is meant).

- Line 257-261: did the meat inspection procedure change in Ireland at some point though?

- Lines 276-287: discussion of the observed effects of age and sex is interesting. Did the authors consider a potential interaction between these covariates?

- Lines 293-295: are these lymph nodes from inconclusive cattle counted as “submitted” in the database? I guess this is an important source of error since for the rest submission will always be linked to the presence of lesions while it wouldn’t be the case for the inconclusive cattle. I would suggest to declare this more openly as a limitation of the study that – hopefully – affects a minor proportion of animals in the database (less than 20k animals in the whole database). Would these animals be preferentially submitted to certain slaughterhouse?

- Lines 302-305: this is a very long sentence that doesn’t read very easily, suggest to break it in two sentences and revise it.

- Lines 323-332: this seems more an introductory section, I would suggest to remove it (or relocate it in the introduction).

- Are (most of) the factories the same as in the previous papers by Frankena et al and Olea-Popelka et al.? If so, and is possible for the authors, it would be interesting to comment on whether factories remain in the same ranking (or quartile) over time.

- A minor suggestion is to remove the second decimal in numbers above 10% (e.g., replace 40.72% by 40.7%, since with large numbers it seems unnecessary). Also, I think sentences should not start with a number (e.g., line 159, start with e.g. “A total of 13,337” instead of by “13,337” directly after the full stop).

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: No

**********

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

15 Jul 2022

We appreciate very much the time and effort the reviewers have taken to read and comment on our manuscript. The comments and suggestion were very helpful to improve the paper and below you can find all our responses with indications to where we have changed the manuscript. We hope that our answers to the queries are satisfactory and that the revised version is now acceptable for publication.

On behalf of all authors,

Rischi Robinson Male Here

(corresponding author)

Reviewer #1: Main considerations

The submitted manuscript reports a study on the submission rate of potentially bovine TB infected lesions from cattle slaughterhouses in Ireland, and on the subsequential bTB confirmation rates of these samples. The study aims to compare the submission and confirmation rates (they are called “risks” in the manuscript, see last point of the main considerations) across different slaughterhouses, and evaluate the risk factor which can have an influence on these rates. In particular, they compare the odds ratio of each slaughterhouse in an univariable model and with a multivariable model, including the number potential factors that might play a role in determining the slaughterhouses confirmation rates. While the subject is of interest there are some pitfalls which undermine the robustness and the depth of these findings.

First, I think there is a confusion between the concepts of “risk” and “rate”. As it stands, I found it very odd to consider that the chance of a sample to be submitted to the lab as a “risk” (same for “confirmation”). I’d suggest to change these to “submission rate” and “confirmation rate”.

AU: by definition a rate has a dimension, e.g. per time unit, which clearly is not the case here. Here, risk is meaning ‘probability’. The description of submission and confirmation risk are stated in lines 101 and 103 as the proportion of slaughtered cows from which a sample has been submitted and the proportion of submitted samples that were confirmed. The same terminology has been used in our previous papers [1,2] and also in others [3] and we would like to stick to this, also for consistency reasons. Perhaps the reviewer is confused by the presentation as ‘per 10,000 animals’ (line 270 - 271) which is just a matter of scaling.

A crucial but neglected aspect is the sampled population spatial distribution, i.e. where the slaughterhouses get the cattle from. Since bTB is not homogeneously spread in Ireland, the location where becomes very important. This is exacerbated by the wildlife interface spread, which makes some areas widely different to others in term of potential bTB prevalence. By neglecting so, it corresponds to assume that all the cattle are homogeneously mixed in Ireland, and that bTB is homogeneous as well, or, that all the slaughterhouses get the cattle form the same exact source. Including cattle geographical factors (e.g. farm coordinates, local TB rates) would be important for the validity of this study, otherwise comparing the slaughterhouses submission and confirmation rates does not provide much information.

AU: this is very valid comment. For privacy, and in compliance with GDPR a data sharing agreement between the data owners (DAFM) and data processors (WUR) for this particular project, factories and herds were anonymised. Therefore, we only had access to the county the cow was send from to the slaughterhouse. However, 73% of the slaughtered animals have moved between herds as the herd of birth is not the same as the herd it was culled from (see table 3, animal movement). Next to that 25 of 31 factories have slaughtered animals from at least 20 (of 26) counties. So it is rather impossible to further elaborate on the spatial distribution as animals are not slaughtered as locally as one would expect and most animals have been in at least 2 herds. Moreover, the herd’s BTB history has been taken into account as covariate in the model and will represent a large part of the spatial heterogeneity, as recrudescence is a characteristic of herd-level bTB epidemiology in Ireland.

Lastly, we remind the reviewer that we are modelling animals with factory lesions, these are animals that did not fail an antemortem test but were found to have evidence of infection post-mortem. Therefore, we expect that the local area prevalence in cattle or wildlife may not have as a significant effect on the sensitivity of the test that within-herd dynamics (see e.g. [4]).

Similarly, no analysis was performed ex-post about what might cause different rates of submission and confirmation in the slaughterhouses. Do ones with similar rates cluster in space, for the type of source herds or else? It would be also interesting comparing the location of the slaughterhouses in relation to the local bTB prevalence, but I understand that location information might be subject to privacy restrictions. Nothing prevent them to be analysed, even if they are not shown on a map.

AU: Indeed variation between factories is interesting, but the reviewer is correct: the slaughterhouses were anonymised and we do not know their location. See also above where it is explained that many animals are not slaughtered locally as slaughterhouses receive animals originating from many counties.

Starting by saying that I am not a statistician, I don’t understand the logic behind the model usage here. In my understanding univariable models are run with a single explanatory variable at a time, and then the most significant ones are run in a multivariable model. At first my best guess was that the univariable model included only the slaughterhouse (as a fixed effect), and that the multivariable model included all the variable in table 1 as fixed effects, and the slaughterhouses as a random effect. However, I realised that the “univariable” model included many variables indeed (see Table S4 and S5). Was the univariable model just a multivariable model but run independently for each slaughterhouse? Or was the slaughtrhouses factor just neglected in the "univariable" model? If this is the case, was the comparison between “univariable” and “multivariable” just a methodological aspect?

AU: the primary factor of interest in this study is the factory and in the univariable model only Factory_id was included as fixed effect (i.e. all factories) and herd_id as random effect. In the multivariable model all potential covariates were added to the univariable model to see if factory rankings would change, while controlling for these herd-level characteristics of the pool of herds contributing to each factory.

The difficulty in understanding what was done is also caused by the absence of any model formula. Even if these are well known methods, reporting a rigorous mathematical formula would help the readers (and the reviewers) in understanding what was done. Finally, no model metric was reported (AIC, BIC). Were all models performing in a similar way? How well these models explained the observations?

AU: The formula for the logistic regression formula is actually quite straightforward. However we have added it, see line 157.

The AIC (‘the smaller the better’) is especially used to find the most parsimonious model which was not the goal of the study. The goal of the paper was to rank factories, not to find the best fitting model, however, we deemed it necessary to investigate (and remove) potential confounding of the factory estimates by covariates.

Finally, I suggest a thorough read of the manuscript before potential resubmissions.

AU: thank you for this advice, we did.

Minor comments

General: the term “factory lesions” sounds odd, it seems like the lesions happened at the factory. I can live with that, but maybe “internal lesions” or something else could be more straightforward for the readers.

AU: factory lesion is a quite common term in TB literature where slaughterhouses are involved (e.g. [4-6]); it means ‘lesions detected in the factory during the slaughter process from antemortem test negative animals’ and is equivalent to Lesions at Routine Slaughter (LRS; e.g. [3,7] which is primarily used in the UK. We have clarified this in line 65 - 66.

Abstract: it would be more beneficial for the readers to have some more background and broader conclusions and considerations in this section, rather than delving into the details of the results.

AU: While we appreciate the reviewer may have read the paper to their taste, we feel that our presentation within the abstract is appropriate for the subject matter and stakeholder audience likely to read or cite this work. Line 14-18 contextualise the work in terms of bovine TB programme within Ireland; line 19-39 outline our approach and major findings; and lines 40-44 present a conclusion. We feel that readers will be informed primarily by the data, more discursive information is presented in the body of the paper for the reader to explore.

Line 49: in Europe only?

AU: the stated reference [2] is about Europe, so we cannot generalise it based on this reference.

Line 56: remove “thus”.

AU: it has been removed.

Line 58: I know that the presence of bTB deer is still controversial, but it is worth mentioning (see Crispell et al., 2020).

AU: thank you, we added the reference, see line 60.

Line 65: the badger culling program?

AU: no, it is culling of cattle; the SICTT targets are bovines which are eliminated if they turn out to be reactors.

Line 81: why an interval is reported here? Is it the percentage of breakdowns per year?

AU: yes the reviewer is correct, the interval shows the variation in the breakdowns per year, sentence adapted (line 84).

Lines 80-83: would it be possible to report all three data in the same format, like all number or percentages (better) of breakdowns disclosed? Otherwise they are difficult to compare.

AU: unfortunately the percentage cannot be deduced from reference [4]. Percentages are more meaningful than numbers as denominators might change considerably over years.

Line 92: remove “the” in front of Ireland.

AU: it has been removed.

Lines 93-94: confusing, please just report the range.

AU: we adapted the text in line 105 – 106.

Lines 99-105: these sentences are convoluted and hard to read.

AU: we have rephrased the paragraph (lines 112-119). Hopefully it is now more clear.

Line 113: is DOD because of “date of death”?

AU: indeed, it has been replaced to date of death (line 127).

Line 127: slaughtered in the same herd or sent to slaughter from the birth herd?

AU: thank you for the suggestion, it has been replaced (line 141).

Line 137: which distribution was calculated? And where is it reported?

AU: it is the frequency distribution. We have adapted it in the text and the distribution is reported in the result (Table 2).

Table 1: are slaughterhouses Z, ZA, ZB, ZC, ZD, ZE related or there are just not enough letters? Did you consider to switch to number or a code instead (S01, S01 or any)?

AU: thank you, we have recoded the factories S01 to S31 in Table 2.

Table 1 and Figure 1: are the slaughterhouses in random order? Would it be more valuable for the readers if they were ranked according to number of slaughtered animals (or some other metric), instead?

AU: the order in Table 2 is in alphabetic order of the (anonymised) slaughterhouse code in the original dataset and is now ordered by magnitude of submission risk. The order in figure 1 and 2 is based on OR (highest to lowest).

Table 1: how solid is the inclusion of slaughterhouse T in the study, given only 2 samples submitted to the lab?

AU: indeed, it seems not to be very solid, but factory T (now S31) fulfilled the inclusion criteria (line 173); it also underlines the large variation between factories.

Lines 173-174: confusing, please rephrase.

AU: we have rephrased (line 193-194).

Line 180: “these animals” are the ones with at least one inconclusive test?

AU:. yes the reviewer is correct. They were inconclusive animals.

Line 185: were “higher”?

AU: ”increased” has been replaced with “higher” (line 205).

Line 186: how is the 3-year threshold being selected? Does it make a difference if the 2 or 4 years would be selected to defined the past bTB history? This sensitivity analysis would be material for supplementary information.

AU: data on how many years a herd has been TB free is not available so we cannot categorise it differently.

Line 224: “lower”?

AU: it has been replaced.

Lines 253-261: I struggle to understand the point of this paragraph, in particular 257-261. Why the VOI procedure should make any difference, if it’s not in place in Ireland?

AU: thank you, we have removed line 257 – 261.

Lines 271-275: were these factors addressed in this analysis?

AU: unfortunately, there is no data available for these factors. These factors might explain part of the variation between factories, probably these factors could be investigated in future studies.

Lines 284-287: herd type, breed, sex and age are intimately intertwined factors in modern cattle industry. Since they are mostly treated as stand-alone factors, I would suggest to integrate them at least in the discussion, so to provide the readers a clearer picture.

AU: We concur that these characteristics can be intertwined given developed cattle milk and beef industries. In this sense, we have discussed how there are inter-relationships between these animal level characteristics, though there is enough variation to fit these variables as independent factors without adding large variance inflation. We have now acknowledged this in the discussion, see line 301.

Lines 323-332: this paragraph is a repetition from the introduction, rather than discussion material.

AU: thank you, it has been moved to the introduction section.

S1: I might have missed this in the text, but why some factories have been excluded from the study?

AU: the exclusion criteria are mentioned in line 173-175.

Reviewer #2: The manuscript “revisiting the relative effectiveness of slaughterhouses in Ireland to detect tuberculosis lesions in cattle (2014-2018)” updates the analyses carried out in the past regarding the differences between slaughterhouses in Ireland in the rate of submission and laboratory confirmation of tuberculosis-compatible lesions. Given the importance of passive surveillance in the eradication programs for bovine tuberculosis in countries where the disease is at (more or less) reduced levels, the continued evaluation of its performance is an objective worth pursuing, and therefore the topic under study here has scientific merit. The analytical framework used is well established and was used in the past with the same objective, and the interpretation of the results is sound. Still, the addition of some details in terms of methodology and results could help to better understand what was performed, and what the results were.

AU: thank you for the review. We really appreciate it.

Comments:

- Line 21-22: the term “ever inconclusive or severe inconclusive skin test reactor” is not very straightforward, suggest replacing it by “ever reactor in the standard or severe skin test” or similar (more similar to what was used in lines 30-31)

AU: thank you for the suggestion. It has been replaced (line 21-22).

- Line 46-47: I would suggest the authors to refer to bovine tuberculosis as the disease caused by members of the M. tuberculosis complex in bovine rather than the disease caused by M. bovis in any host species (though both definitions are widely used in the literature, I think we would not refer to bovine brucellosis as the infection by B. abortus in sheep, and the former definition is more aligned with current regulations in the EU).

AU: we changed it accordingly. Please see line 47-48.

- Line 51: suggest replacing “in” by “by” (since countries did not gain the OTF status precisely in 2019)

AU: replaced (line 52).

- Lines 94-97: Given that in the papers mentioned in the previous lines (refs 15 and 20) at least some of the covariables mentioned here (age, gender…) were considered in the multivariable models, I would expect that variations in their distributions depending on the slaughterhouse should not explain the difference in submission risk per slaughterhouse (that is the point of considering them in the analysis, isn’t it?).

AU: yes, the reviewer is correct, covariates were added in order to get adjusted estimates for the factories and in that way the submission risk does not depend anymore on the differences in type of animals slaughtered.

- Line 113: I think gender is a trait only applicable to humans, and sex should be used here (as e.g. used in Table 1).

AU: corrected (line 108 and 127).

- Line 114: for purchased animals: did the herd history include all the annual tuberculin testing in all herds where the animal had stayed or only the last one prior to be shipped to the abattoir? (the last testing date is mentioned in line 115 but I wonder if this refers to the whole information on skin testing).

AU: this is a good point. The herd history concerns information on the herd the animal last resided in before slaughter. This was a herd level factor, and therefore did not “follow” the animal, and was included for the potentiality of animal level exposure if residual infection was a problem in the herd. ‘Life long’ metrics of exposure would be a useful factor to explore for a future analysis.

- Data analysis: Herd size is an interesting covariable but I wonder if it makes sense to assume a linear relationship between the (logit of) the probability of submission/confirmation and the number of animals in the herd. Did the authors consider categorizing in some way this variable and see whether it improved (or not) model fit? Although categorizing a continuous variables has its own drawbacks, in terms of interpretation it may be easier to see increase in the risk in “large” vs. “small” herds (just a suggestion). Also, it seems (based on Table S4) it was log-transformed; if so, this should be indicated in the methodology section.

AU: indeed categorising continuous variables should be done on some (bio)logical basis. We ran the models with 4 herd categories based on quartiles. It did not have an effect on factory rankings. Log transformation is now mentioned in M&M (Table 1).

- Data analysis (II): I would be interested in knowing the view of the authors regarding considering factory as a random rather than as a fixed effect. Though I see the point in considering it a fixed effect, given the very large number of categories and the potential objective of characterizing variability across slaughterhouses assuming the belong to some sort of population, coupled with the view of the database as a hierarchical structure (with animals clustered in abattoirs). Note that I don’t think there is anything wrong with the approach used and I am not requesting the authors to change it, but I would like to know if this was considered at some point or if the authors have a strong view against this option.

AU: we did not consider this as an option as we needed to have estimates (odds ratios) in order to rank the factories. In some papers factory is included as random effect but then the question behind the analyses was different, i.e. to get adjusted estimates for specific risk factors, see e.g. [3].

- Line 155: I was surprised to see animals <1 year were excluded but looking at S2 I realize they represent a very small fraction of the population. In other countries it may be not so unusual to slaughter animals below one year (10-11 months-old), so perhaps the fact that animals below 1 year were excluded could be added in the material and methods as an exclusion criterion

AU: thanks for this useful comment, we added it as exclusion criterion, see line 173-175.

- Line 153: suggest to specify that there are four small factories (I think there are four based on line 158, 31 factories after exclusion of small ones).

AU: we have added the number of factories excluded in line 173.

- Lines 161-162: what does this correlation refer to? The crude submission and confirmation risks per factory? This could be perhaps added as a supplementary graph (and although I assume it is considered included in the “descriptive analysis” part on the material and methods, also be specified there as an analysis performed).

AU: indeed it concerns the confirmation risk and submission risk of the factories. A supplementary graph would just show that there is no relation between submission risk and confirmation risk as the correlation is nearly zero. We now have mentioned the correlation in M&M (line 150-151).

- Line 175: suggest replacing “ranged” by “ranging”

AU: it has been replaced (line 195).

- Modeling outcomes: I would suggest to start both submission and confirmation sections by indicating which variables (all considered, I pressume?) were included in the final model.

AU: we adapted the text. See lines 211-212 and 237-238.

-Also, I would strongly suggest to add a table in the main text with the adjusted OR coefficients (along with some of the descriptives included in Table S2?) for these covariables (something similar to Table 1 in the paper by Frankena et al). Though this is included in table S4, I think it is very useful information that readers would appreciate having in the main text (and could be easily integrated in Table 1, that currently conveys only limited information).

AU: thank you for the useful suggestion. The table with descriptive and OR of risk factor used in the analysis has been added to the text (Table 3).

- Line 191-192 and 216-217: given that factory-specific ORs are calculated based on the difference with the baseline abattoir, I am not sure how to interpret these sentences (differences between factories become larger once covariates were considered? But this would be only considering the reference category, wouldn’t it? Perhaps other differences decrease?)

AU: Thank you very much for this remark. Indeed the difference is relative to the reference category. We adjusted the text to avoid confusion .

- Line 251: suggest replacing “difference” by “increase” (since I think this is what is meant).

AU: replaced (line 275).

- Line 257-261: did the meat inspection procedure change in Ireland at some point though?

AU: thanks for the question, no major changes have been applied.

- Lines 276-287: discussion of the observed effects of age and sex is interesting. Did the authors consider a potential interaction between these covariates?

AU: no, we did not evaluate interactions as the main interest was on the factory ranking.

- Lines 293-295: are these lymph nodes from inconclusive cattle counted as “submitted” in the database?

AU: it depends on the farmer’s decision, sending inconclusive animals to slaughter is not compulsory. A very high % from the ever inconclusive animals is submitted, regardless of the presence of visible lesions. This then drives down the proportion that are confirmed. The opposite is happening with the ever severe inconclusive, where a relatively small proportion is submitted with a very high confirmation %.

I guess this is an important source of error since for the rest submission will always be linked to the presence of lesions while it wouldn’t be the case for the inconclusive cattle. I would suggest to declare this more openly as a limitation of the study that – hopefully – affects a minor proportion of animals in the database (less than 20k animals in the whole database). Would these animals be preferentially submitted to certain slaughterhouse?

AU: thanks for this useful comment. Indeed, it is a minor proportion of the ever inconclusive cattle (11,526/6,578,245 =0.17%, Table 3). We referred to this in the Discussion, see lines 316-320. Further analysis of the data show that many factories process ever inconclusive animals. Thus, these animals are not processed in specific slaughterhouses.

- Lines 302-305: this is a very long sentence that doesn’t read very easily, suggest to break it in two sentences and revise it.

AU: we rephrased the sentence (328-332).

- Lines 323-332: this seems more an introductory section, I would suggest to remove it (or relocate it in the introduction).

AU: thank you, we have relocated the paragraph to the introduction section, see lines 87 – 95.

- Are (most of) the factories the same as in the previous papers by Frankena et al and Olea-Popelka et al.? If so, and is possible for the authors, it would be interesting to comment on whether factories remain in the same ranking (or quartile) over time.

AU: factories are partly the same but as all were anonymised in all studies no comparison can be made; only the average levels and the variation between factories can be evaluated.

- A minor suggestion is to remove the second decimal in numbers above 10% (e.g., replace 40.72% by 40.7%, since with large numbers it seems unnecessary).

AU: in principal the reviewer is correct. However, for consistency we used two decimals for all small and large figures.

Also, I think sentences should not start with a number (e.g., line 159, start with e.g. “A total of 13,337” instead of by “13,337” directly after the full stop).

AU: thank you very much for the useful suggestion, we adapted the text where it occurred.

1. Frankena K, White PW, O'Keeffe J, Costello E, Martin SW, Van Grevenhof I, et al. Quantification of the relative efficiency of factory surveillance in the disclosure of tuberculosis lesions in attested Irish cattle. Vet. Rec. 2007;161(20):679-84. doi: 10.1136/vr.161.20.679.

2. Olea-Popelka FJ, Freeman Z, White P, Costello E, O'Keeffe J, Frankena K, et al. Relative effectiveness of Irish factories in the surveillance of slaughtered cattle for visible lesions of tuberculosis, 2005-2007. Ir. Vet. J. 2012;65(1). doi: 10.1186/2046-0481-65-2.

3. Pascual-Linaza AV, Gordon AW, Stringer LA, Menzies FD. Efficiency of slaughterhouse surveillance for the detection of bovine tuberculosis in cattle in Northern Ireland. Epidemiol. Infect. 2017;145(5):995-1005. doi: 10.1017/S0950268816003095.

4. Byrne AW, Barrett D, Breslin P, Madden JM, O’Keeffe J, Ryan E. Post-mortem surveillance of bovine tuberculosis in Ireland: herd-level variation in the probability of herds disclosed with lesions at routine slaughter to have skin test reactors at follow-up test. Vet. Res. Commun. 2020;44(3-4):131-136. doi: 10.1007/s11259-020-09777-w.

5. Willeberg PW, McAloon CG, Houtsma E, Higgins I, Clegg TA, More SJ. The herd-level sensitivity of abattoir surveillance for bovine tuberculosis: simulating the effects of current and potentially modified meat inspection procedures in Irish cattle. Front. Vet. Sci. 2018;5(82). doi: 10.3389/fvets.2018.00082.

6. Clegg TA, Good M, More SJ. Future risk of bovine tuberculosis recurrence among higher risk herds in Ireland. Prev. Vet. Med. 2015;118(71-79). doi: 10.1016/j.prevetmed.2014.11.013.

7. Shittu A, Clifton-Hadley RS, Ely ER, Upton PU, Downs SH. Factors associated with bovine tuberculosis confirmation rates in suspect lesions found in cattle at routine slaughter in Great Britain, 2003–2008. Prev. Vet. Med. 2013;110(3-4):395-404. doi: 10.1016/j.prevetmed.2013.03.001.

Submitted filename: Response to Reviewers.docx

Click here for additional data file.

7 Sep 2022

9 Sep 2022

We appreciate very much the time and effort the reviewers have taken to read and comment on our manuscript. The comments and suggestion were very helpful to improve the paper and below you can find all our responses (in bold italics) with indications to where we have changed the manuscript. We hope that our answers to the queries are satisfactory and that the revised version is now acceptable for publication.

On behalf of all authors,

Rischi Robinson Male Here

(corresponding author)

Reviewer #1: The effort of the authors made in addressing the syntax and clarity of the manuscript were sound, and the methodology and model explanation are cleaer now. However, they were short in addressing many others concerning aspects concerning terminology (which could have been improved, but it wasn't) and, in particular, giving a broader context and depth to the results, and making the study a more interesting one.

AU: Thank you for your time and consideration. The terminology is in accordance with previous publications on the same topic both from the same group (Frankena, Olea-Popelka) as from other groups (Northern Ireland). Consistency indeed is important and we have the opinion to have done so.

Reviewer #2: The revised version of the manuscript has satisfactorily addressed the points raised in the previous review. I have only a question and a couple of editorial suggestions

- I was a bit confused by one of the replies given to reviewer#1 regarding the purpose of multivariable models: this reply stated “in the univariable model only Factory_id was included as fixed effect (i.e. all factories) and herd_id as random effect”. From table 3 it seems obvious that in fact all available covariates were expressed in univariable models (this makes sense in my view). Based on the answer I assume that these univariable models also included a random effect (I think this also makes sense), and I assumed they did not include the factory (but based on the answer I am not so sure anymore). Perhaps this could be better clarified in lines 154-155 (version with track changes). A suggestion would be “The effect of each risk factor on the FL submission and confirmation risk was analysed with multilevel univariable logistic regression, followed by multilevel mixed effects logistic regression with…”, i.e., adding the multilevel to the univariable part to make evident that random effects were also included in the univariable models. Then, if all univariable models included the factory, perhaps state in that sentence this (since if the model includes e.g. factory and year does not fit the classical definition of a univariable regression model).

AU: Thank you for your comment and suggestion. We apologize and we understand your confusion. The reviewer is correct that all available covariates were expressed in univariable models as shown in Table 3. Also, we want to clarify that factory was not included in these univariable models. In the first rebuttal letter it was stated that herd was included as random effect in the univariable models; however the OR’s and CI’s in Table 3 originated from univariable models without that random effect. Therefore we corrected the univariable results presented in Table 3 and updated the text and figures where necessary, leading to very minor changes. Also we now explicitly mention in M&M that the univariable models included a random effect (line 152).

Lines 234 and 260 (version in track changes): in the captions of figures 3 and 4, replace “referent” by “reference category”.

AU: Thank you for the suggestion. It has been replaced (line 230 and 255).

To the editor: we would like to update the funding disclosure to clarify the funder of the article processing costs. The updated funding disclosure is given below:

“The article processing costs associated with the publication of this study were provided by the Department of Agriculture, Food and Marine, Ireland. RRMH was sponsored by Indonesia Endowment Fund for Education scholarship (LPDP, https://lpdp.kemenkeu.go.id) from the Ministry of Finance, the Republic of Indonesia. The funder had no role in study design, data collection, and analysis, decision to publish, or preparation of the manuscript”.

Submitted filename: Response to Reviewers.docx

Click here for additional data file.

13 Sep 2022

Revisiting the relative effectiveness of slaughterhouses in Ireland to detect tuberculosis lesions in cattle (2014-2018)

PONE-D-22-10323R2

Dear Dr. Male Here,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Rebecca Lee Smith, D.V.M., M.S., Ph.D.

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

Reviewers' comments:

28 Sep 2022

PONE-D-22-10323R2

Revisiting the relative effectiveness of slaughterhouses in Ireland to detect tuberculosis lesions in cattle (2014-2018)

Dear Dr. Male Here:

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

If we can help with anything else, please email us at plosone@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. Rebecca Lee Smith

Academic Editor

PLOS ONE