Rodrigues da Costa et al. Porcine Health Management (2019) 5:4 https://doi.org/10.1186/s40813-018-0113-6 RESEARCH Open Access Using the Biocheck.UGent™ scoring tool in Irish farrow-to-finish pig farms: assessing biosecurity and its relation to productive performance Maria Rodrigues da Costa1,2* , Josep Gasa2, Julia Adriana Calderón Díaz1, Merel Postma3, Jeroen Dewulf3, Gerard McCutcheon4 and Edgar Garcia Manzanilla1,5 Abstract Background: Biosecurity is one of the main factors affecting disease occurrence and antimicrobial use, and it is associated with performance in pig production. However, the importance of specific measures could vary depending on the (national) context. The aim of this study was to describe the biosecurity status in a cohort of Irish pig farms, to investigate which of those biosecurity aspects are more relevant by using the Biocheck.UGent™ scoring system, and to study the impact of such aspects on farm performance. Results: External biosecurity score was high compared to most countries due to the characteristics of the Irish pig sector (i.e. purchasing only semen and breeding gilts on farm). The internal biosecurity score was lower and had greater variability among farms than other EU countries. Using multivariable linear regression, the biosecurity practices explained 8, 23, and 16% of variability in piglet mortality, finisher mortality, and average daily gain, respectively. Three clusters of farms were defined based on their biosecurity scores (0 to 100) using principal components and hierarchical clustering analysis. Scores for clusters 1, 2 and 3 were (mean ± SD) 38 ± 7.6, 61 ± 7.0 and 66 ± 9.8 for internal and 73 ± 5.1, 74 ± 5.3 and 86 ± 4.5 for external biosecurity. Cluster 3 had lower piglet mortality (P = 0.022) and higher average daily gain (P = 0.037) when compared to cluster 2. Conclusions: Irish farms follow European tendencies with internal biosecurity posing as the biggest liability. Our results suggest that practices related to the environment and region, feed, water and equipment supply, and the management of the different stages, need to be addressed in lower performing farms to improve productive performance. Further studies on the economic impact of these biosecurity practices including complementary data on herd health, gilt rearing, piglet management, vaccination and feeding strategies are needed. Keywords: Biosecurity, Efficiency, Ireland, Productive performance, Swine production Background as key inputs for productivity, and highlighted that the Pig production is the third biggest agricultural activity in biggest challenges in the Irish setting are the control of Ireland [1], with a self-sufficiency of 219% [2]. In 2016, endemic diseases, the prevention of introduction of there were approximately 150,000 breeding sows, produ- exotic diseases, and the reduction of the use of antimi- cing an estimate of 4,000,000 pigs per year [2]. The Irish crobials, among others [1]. In the last few years, biose- pig sector has identified animal health and management curity practices have been widely discussed. On-farm biosecurity protects farms from disease by preventing * Correspondence: Maria.costa@teagasc.ie pathogenic agents to enter (external biosecurity) or 1 Pig Development Department, Teagasc - Animal & Grassland Research and spreading once inside the farm (internal biosecurity, Innovation Centre, Moorepark, Fermoy, Co. Cork, Ireland 2 Departament de Ciència Animal i dels Aliments, Facultat de Veterinaria, which can also overlap with management). Good biose- Universitat Autònoma de Barcelona, Bellaterra, 08193 Barcelona, Spain curity practices were related to improved performance, Full list of author information is available at the end of the article © The Author(s). 2019 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Rodrigues da Costa et al. Porcine Health Management (2019) 5:4 Page 2 of 9 better financial return for farmers [3, 4], and to a low Laanen et al. [5]. Briefly, the questionnaire has 109 closed use of antimicrobials [5, 6]. Moreover, few studies provide questions grouped in 12 categories corresponding to ei- quantitative data effectively linking production perform- ther external (six categories) or internal biosecurity (six ance to biosecurity [5, 7, 8]. The Biocheck.UGent™ scoring categories). Each category assesses several practices and system developed by Gent University [9] assesses biose- its score was given in a rank from 0 (worst scenario) to curity using a risk assessment approach and it has been 100 (best scenario). Each question had a fixed score. Ex- successfully applied in several EU countries [3, 10, 11]. ternal and internal biosecurity scores were computed as a Postma et al. [3] showed that biosecurity has moderate weighted average of the scores achieved in the corre- correlations to production performance in four European sponding categories. Overall biosecurity was computed as countries (Belgium, France, Germany, Sweden) and con- the average of external and internal biosecurity scores. A cluded that biosecurity practices vary with the country. paper copy of the questionnaire was completed at the This indicates that characterization and contextualization farm with both farmer and researcher and the results were of the pig sector in each national setting are important to transcribed to the Biocheck.UGent™ online database and understand the connection between biosecurity and per- Microsoft Office Excel format. The final scores for each formance [11]. For research purposes, this contextualization biosecurity category were obtained for each farm and were is difficult when the methods used across countries are not used for the analysis. the same. However, record keeping and benchmarking within and between countries are necessary tools for an ef- Performance data ficient and competitive sector. In this study, we aimed to Performance data for the year 2016 were retrieved from describe biosecurity status in Irish pig farms, to investigate the Teagasc ePM database for all 58 farms included in which biosecurity aspects were more relevant by using the the study. ePM data was collected on farms every tri- Biocheck.UGent™ scoring system, and to study the impact mester with the support of Teagasc advisors and collated of such aspects on farm performance. into a single database. This information was used to pro- duce the yearly National Pig Herd Performance Report Methods [13], and for different international comparisons like the Fifty-eight Irish pig farms were scored between February InterPIG report [14]. The productive performance indi- and May 2016 using the biosecurity scoring system Bio- cators selected for analysis in the present study were pig- check.UGent™. Performance data for 2016 for these let (pre-weaning) and finisher mortality (%), number of farms were retrieved from Teagasc e-Profit Monitor pigs produced per sow per year, average daily gain (ePM) – a national herd monitoring system – and the (ADG, g/day) corresponding to the period from weaning effect of the biosecurity practices on selected productive to finish, and feed conversion ratio (FCR), corresponding performance indicators was estimated. to the same period. Farm selection Statistical analysis The Teagasc ePM is a herd monitoring system available All statistical procedures were performed in R version on a voluntary basis to all the farmers in the Republic of 3.4.4 (Vienna, Austria, [15]). Alpha level for significance Ireland. In 2016, it included 129 pig herds representing and tendency were 0.05 and 0.10, respectively. Differ- over 96,000 sows or 65% of the national commercial sow ences in productive performance between the study herd. All the farmers providing data to the ePM were of- sample and the Teagasc ePM population were tested by fered the biosecurity assessment of their farrow-to-finish means of independent samples t test (means) and F-tests farms using Biocheck.UGent™ and 58 farmers partici- (variance). The effect of biosecurity on productive per- pated voluntarily. Farms were recruited through the Tea- formance was estimated through multivariable linear gasc advisory service and represent approximately 29% models. Productive performance indicators were used as of the national commercial sow population. dependent variables and basic farm characteristics not included in the biosecurity score (number of sows, years Biosecurity assessment of experience of farm managers, number of workers, age Four researchers visited farms and interviewed farmers of the oldest building in which pigs were kept, and age to complete the Biocheck.UGent™ questionnaire. All in- of the youngest building in which pigs were kept), and terviewers were trained to conduct the questionnaire. biosecurity scores were used as independent variables or The training provides the criteria to frame the questions predictors. First, Spearman Rank correlations were cal- and provides examples on how to complete the question- culated between all the independent variables to detect naire given different scenarios to reduce inter-observer collinearity. Then, for each performance indicator, a variability. A detailed description of the questionnaire and multivariable linear model was fitted with predictors se- its scores was explained by Backhans et al. [12] and lected from the farm characteristics, and the biosecurity Rodrigues da Costa et al. Porcine Health Management (2019) 5:4 Page 3 of 9 categories’ scores. When fitting the model for piglet oldest farm buildings in which pigs were kept were on mortality, biosecurity scores related to the nursery and average 35.3 ± 25.46 years old, this figure includes a farm finishing unit management were left out. A forward re- which was 210 years old (range = 5–210). The second gression approach was used to improve the models fitted oldest farm was 60 years old. The youngest buildings (ols_step_forward function from the olsrr package in R were on average 3.9 ± 5.14 years old with some farms [16]), using a cut-off value of 0.15 for predictor retention reporting to be building new accommodation at the time in the model. Predictors collinearity was further checked (range = 0–25). In this study, 34.5% of the farms re- using Variance Inflation Criterion (VIF) from the R ported keeping other animals (cattle) for commercial package rms [17]. Finally, for each model, residuals’ nor- purposes on the farm grounds. Of those farms, 50% kept mality was visually assessed. A simple linear model fit- beef and 55% kept dairy cattle. One farm kept sheep be- ting internal biosecurity scores from external scores was sides beef. Farm characteristics and herd productive also done. performance (Table 1) did not differ between farms To identify the most relevant biosecurity aspects in keeping other animals for commercial purposes and Irish pig farms, a Principal Component Analysis (PCA) farms keeping only pigs, except for ADG (740 ± 57.1 was also performed on the biosecurity categories and the vs. 685 ± 56.3 g/day, respectively; P < 0.001). 2 main dimensions for the principal components were described. After this, Hierarchical Clustering Analysis Farm productive performance was used to group farms according to their similarities The average herd size of the farms included in the study regarding their internal and external biosecurity prac- was 754 sows (range = 113–2479). All the values obtained tices. Biosecurity scores and productive performance for for productive performance indicators were in the each cluster were compared by ANOVA followed by inter-quartile range for the InterPIG report [14], except Tukey’s correction. for piglet mortality which is on the lowest 25%. For ADG, in Ireland it included the period from weaning (29 ± 3.7 d; Results 7.1 ± 0.46 kg) to finish (107 ± 5.2 kg live-weight). Piglet Farm characteristics and finisher mortality showed great variability across Table 1 summarizes the characteristics of the 58 pig farms with a coefficient of variation (CV) of 28.7 and farms included in this study. The farms employed farm 44.7%, respectively. The number of pigs produced per sow managers with an average of 26.8 ± 10.14 years of experi- per year, ADG (g/day), and FCR showed less than 10% ence. The number of sows (hereinafter, average herd variability across farms (CV = 8.7, 8.8, and 5.9%, respect- size) was strongly correlated to the number of workers ively). Between the study sample and the ePM population, on farm (rs = 0.92, P < 0.001) with a ratio of one worker differences were found only in the variance of finisher per 154 ± 34.1 sows (range = 55–210). Although the mortality (%) and ADG (g/day) (P > 0.001 and P = 0.037, Table 1 Description of the sample of 58 Irish farrow-to-finish pig farms used in the 2016 biosecurity assessment and comparison to the Teagasc database (ePM) population (n = 129) Item ePM mean ± SD Study samplea mean ± SD median (Range) Farm characteristics No. of sows 726 ± 610.8 754 ± 554.9 639 (113–2479) Experience of farm manager, years – 26.8 ± 10.14 28.0 (5.0–50.0) Number of workers – 4.9 ± 3.65 4.0 (1.0–16.0) Age of the oldest building, years – 35.3 ± 25.46 32.5 (5.0–210.0) Age of the youngest building, years – 3.9 ± 5.14 3.0 (0.0–25.0) Herd productive performance No. of pigs produced per sow per year 25.7 ± 2.30 26.0 ± 2.27 25.8 (18.0–31.2) Piglet mortality, % 10.5 ± 2.80 10.3 ± 2.70 9.8 (5.1–16.3) Finisher mortality, % 2.4 ± 1.47 2.2 ± 0.97 2.0 (0.8–5.1) Average daily gainb, g/day 703 ± 79.8 704 ± 62.0 699 (554–856) b Feed conversion ratio 2.41 ± 0.171 2.38 ± 0.144 2.36 (2.01–2.78) a Farm characteristics retrieved from the Biocheck.UGent™ scoring tool which was applied to 58 Irish farrow-to-finish pig farms from February to May 2016. The correspondent herd productive performance was retrieved from the Teagasc ePM for the year 2016 b Average daily gain and Feed conversion ratio correspond to the period from weaning to finish Rodrigues da Costa et al. Porcine Health Management (2019) 5:4 Page 4 of 9 respectively), but not in their means. No other differences low variability (CV = 5.1%). Table 3 summarizes the regarding means or variance were found (P > 0.05) across models selected. average herd size, piglet mortality (%), number of pigs per The model selected for piglet mortality (%) explained sow per year, or FCR. 8% of the variability. There was an increase in mortality with age of the youngest building in which pigs were kept (P < 0.001), and a tendency for it to decrease in Biosecurity scores farms with better scores in the biosecurity category re- The response rate was 100% for all the questions in the ferring to feed, water, and equipment supply (P = 0.079). questionnaire. The results of the biosecurity assessment The model for finisher mortality (%) explained 23% of the are presented in Table 2. The overall biosecurity score was variability. Mortality increased with the average herd size 68.3 ± 9.52. Total external biosecurity scored higher than (P < 0.001) and decreased with good disease management internal biosecurity (P < 0.001) and its practices were ap- scores (P = 0.028). High scores in the categories environ- plied consistently across farms (CV = 9.8%). The highest ment and region and in nursery unit management tended score in this category was achieved in the category pur- to be related to higher mortalities (P = 0.059 and P = 0.050, chase of animals and semen (98.8 ± 5.05, range = 70–100). respectively). Good measures between compartments and The lowest score in this category was in the feed, water, use of equipment seemed to decrease finisher mortality al- and equipment supply (54.5 ± 14.57). Regarding internal though this was not statistically significant (P = 0.126). biosecurity, disease management scored the highest with The model for ADG (g/day) explained 16% of the 82.4 ± 21.55, and cleaning and disinfection obtained the variability. It decreased in large farms (No. of sows, lowest score (42.0 ± 27.25) with 12.1% of the farms not ap- P = 0.043) and with the experience of the farm man- plying any of these practices (score 0). ager (P = 0.029). Good practices in disease manage- ment improved ADG (P = 0.039). Effect of biosecurity scores and farm characteristics on productive performance Relationship between internal biosecurity and external The number of workers was left out of the predictors due biosecurity to collinearity with average herd size (rs = 0.92, P < 0.001). Around 20% of the variability in internal biosecurity (ad- Among the biosecurity categories, the purchase of animals justed R2 = 0.20, P < 0.001) could be explained by the and semen was also left out of the predictors due to its scores obtained in external biosecurity: Table 2 Biosecurity scores (Biocheck.UGent™) for the different categories of internal and external biosecurity in 58 farrow-to-finish Irish pig farms Mean SD Median Min Max External biosecurity scorea 78.7 7.75 79.0 62.0 94.0 Purchase of animals and semen 98.8 5.05 100.0 70.0 100.0 Transport of animals, removal of manure and dead animals 80.1 11.26 83.0 43.0 96.0 Feed, water, and equipment supply 54.5 14.57 53.0 10.0 80.0 Personnel and visitors 73.9 18.61 76.0 24.0 100.0 Vermin and bird control 68.3 19.84 70.0 30.0 100.0 Environment and region 79.5 23.35 80.0 20.0 100.0 Internal biosecurity scorea 57.4 14.16 60.0 29.0 80.0 Disease management 82.4 21.55 80.0 20.0 100.0 Farrowing and suckling period management 53.6 18.75 57.0 7.0 86.0 Nursery unit management 63.5 16.11 64.0 36.0 100.0 Fattening unit management 72.7 22.12 79.0 21.0 93.0 Measures between compartments and use of equipment 50.0 16.16 50.0 21.0 86.0 Cleaning and disinfection 42.0 27.25 40.5 0 95.0 Overall biosecurity score 68.3 9.52 70.0 47.0 87.0 a Biosecurity scores are computed from the practices assessed in each category. Category scores are given in a rank from 0 (worst scenario) to 100 (best scenario). External and internal biosecurity scores correspond to the average of the scores obtained in the corresponding categories. The overall biosecurity corresponds to the average between the external biosecurity score and the internal biosecurity score Rodrigues da Costa et al. Porcine Health Management (2019) 5:4 Page 5 of 9 Table 3 Multivariable linear regression modelling of herd productive performancea Outcome Predictor Estimate SE P-value Piglet mortality, % Intercept 12.04 1.334 < 0.001 Adjusted R2 = 0.08 Age of the youngest building, years 0.13 0.066 0.067 P = 0.039 Score for feed, water, and equipment supply −0.04 0.023 0.079 Finisher mortality, % Intercept 1.50 0.683 0.032 Adjusted R2 = 0.23 No. of sows [per 100 sows] 0.8 0.21 < 0.001 P = 0.002 Score for disease management −0.01 0.006 0.028 Score for environment and region 0.01 0.005 0.059 Score for nursery unit management 0.02 0.008 0.050 Score for measures between compartments and use of equipment −0.01 0.008 0.126 ADG, g/day Intercept 706.27 37.734 < 0.001 Adjusted R2 = 0.16 No. of sows −0.03 0.0133 0.043 P = 0.006 Experience of farm manager, years −1.65 0.734 0.029 Score for disease management 0.73 0.343 0.039 a Each productive performance indicator (piglet mortality (%), finisher mortality (%), number of pigs per sow per year, ADG (g/day), and FCR) was modelled from herd characteristics and biosecurity scores (categories), presented in Tables 1 and 2, respectively. The table presents the final models after a forward regression approach with a cut-off value of 0.15 for predictor retention. The models fitting the number of pigs per sow per year and FCR were not significant (overall F-test with P = 0.067 and P = 0.075, respectively) Internal biosecurity ¼ −8:434 þ 0:836 compartmentalization, transport of animals and removal  External biosecurity of manure and dead animals, and management of the different stages in dimension 1 (mainly internal biose- curity categories); and the environment and region, feed, Farm clusters based on biosecurity practices water, and equipment, management of the different The first two dimensions of the PCA of the farms de- stages, and personnel and visitors in the dimension 2 pending on their biosecurity practices accounted for (mainly external biosecurity categories). Three clusters 47% of variability. Dimension 1 accounted for 33.6% of of farms were identified based on their similarities in variability and was mainly explained (66.1%) by internal biosecurity practices (Fig. 1). The average internal biosecurity practices. Dimension 2 accounted for 13.4% biosecurity score in cluster 1, 2, and 3 was (mean ± SD) of variability and was mainly linked to external biosecur- 38.4 ± 7.6, 61.4 ± 6.99, and 66.3 ± 9.81, respectively. This ity (68.4%). The main categories contributing to the score differed statistically between cluster 1 and cluster clustering of the farms were: cleaning and disinfection, 2 or 3 (P < 0.001). The average external biosecurity score Fig. 1 Clusters of farms grouped according to their biosecurity scores in external and internal biosecurity categories. Legend: Dim1 – Dimension 1; Dim2 – Dimension 2. A Principal Components Analysis followed by Hierarchical Clustering Analysis grouped farms according to their scores in external and internal biosecurity practices. Dimension 1 was mainly related with internal biosecurity and dimension 2 was mainly related with external biosecurity. The three clusters identified group farms with low internal biosecurity and high external biosecurity (cluster 1), average internal and external biosecurity (cluster 2), and high internal and external biosecurity (cluster 3) Rodrigues da Costa et al. Porcine Health Management (2019) 5:4 Page 6 of 9 in cluster 1, 2, and 3 was 73.2 ± 5.12, 74.4 ± 5.33, and Internal biosecurity scores showed a lack of compliance 86.1 ± 4.47. This score differed statistically between clus- with cleaning protocols and compartmentalization within ter 3 and cluster 1 or 2 (P < 0.001). No other differences the farm. The structure of the farms in Ireland (farrow- were found between clusters in regard to internal and ex- to-finish) and the heterogeneity of the facilities (age of the ternal biosecurity scores. The productive performance in- buildings) contribute to the variability seen in internal bio- dicators for each cluster of farms are presented in Fig. 2. security practices. Many Irish farms have grown in size by Cluster 2 and cluster 3 were different for piglet mor- adding new buildings to older, but still functional, facil- tality (11.6 ± 2.84% vs. 9.4 ± 2.39%, P = 0.022), and ities. This heterogeneity makes standardization of proto- ADG (679 ± 68.2 g/day vs. 726 ± 58.3 g/day, P = 0.037). cols to control diseases like PRRS difficult. It is likely that Cluster 2 and 3 tended to be different in the number of pigs farmers do not valorise the pertinence of internal biose- per sow per year (25.2 ± 1.71 vs. 26.8 ± 2.08, P = 0.057). curity. Casal et al. [18] states that farmers are likely to im- Finisher mortality and FCR did not differ between clusters plement biosecurity measures they perceive as important. (P = 0.956 and P = 0.131, respectively). However, the awareness towards biosecurity has tradition- ally been focused on external biosecurity by farmers to Discussion avoid those diseases not endemic to their farms. In recent The aim of this study was to describe biosecurity years, the key importance of internal biosecurity practices practices among Irish pig farms and their relationship to reduce disease and improve profitability has resurged with productive performance. The overall biosecurity and gained new strength. In any case, internal biosecurity scores agreed to what has been described in Belgium, was positively correlated to external biosecurity in accord- France, Germany, Sweden, and Denmark [3, 11]. Ex- ance with other studies highlighting the inter-relationship ternal biosecurity was higher in Ireland than in other between both aspects of biosecurity [3, 10–12]. countries, except for Denmark whose pig sector is fo- The multivariable model for pig mortality explained cused on high health with strict biosecurity practices only 8% of the variability. The age of the buildings was [11]. The Irish pig sector includes mostly closed directly related to mortality. Although 79% of the farms herds, resulting in less animal movements with 94.5% had built new housing for pigs within the previous 5 of the farms reported to buy only semen from genetic years, several farms had their latest renovation 10 or companies and not purchasing gilts. In other coun- 15 years ago. Piglet mortality was also associated to tries, gilts are purchased and quarantined on farm be- poor biosecurity on feed, water, and equipment supply. fore introducing them to the herd [3, 18, 19]. This is As shown in the case of porcine epidemic diarrhoea, a risk for external biosecurity not existing in Ireland. these supplies increase the risk of introduction of new However, rearing gilts on site may have negative effects on diseases, which can be linked to higher mortality. Sur- internal biosecurity that are not included in this study. prisingly, the farrowing unit management was not Fig. 2 Boxplots of productive performance indicators (with mean ± SD) across farm clusters of farms grouped according to their biosecurity scores in external and internal biosecurity categories. Legend: ADG (g/day) – Average daily gain, FCR – Feed conversion ratio. The clusters represent farms with similar biosecurity scores in external and internal categories. Cluster 1 groups farms with low internal biosecurity and high external biosecurity, cluster 2 groups farms with average external and internal biosecurity, and cluster 3 groups farms with high external and internal biosecurity scores. The productive performance of the farms in each cluster is presented above. ANOVA tests followed by Tukey’s correction were used to test differences in productive performance across those clusters Rodrigues da Costa et al. Porcine Health Management (2019) 5:4 Page 7 of 9 retained in the piglet mortality model. This manage- practices. Indeed, the Biocheck.UGent™ was, as many ment, as measured by the Biocheck.UGent™, focuses on other biosecurity assessment tools, designed by expert cross-fostering practices, disinfection of materials be- panels using experience and logical reasoning but not tween litters, and castration protocols. Other factors scientific validation to support biosecurity practices [7]. such as sow management, farrowing supervision, colos- Thus it lacks the baseline factors impacting on perform- trum intake, split suckling and training of staff [20, 21] ance such as herd health status, genetics, use of antimi- may have a greater impact on piglet mortality than the crobials and vaccinations, and feeding practices. An practices captured in the Biocheck.UGent™. alternative to this bias would be to model performance The model for finisher mortality explained 23% of the using the practices assessed (individual questions) in- variability. Bigger farms had higher finisher mortality. stead of aggregated scores. Finally, some of these tools Although in our data the bigger the farm, the higher the were designed to address certain pathogens (i.e. PRRSv), number of workers, we suspect that bigger farms may not necessarily providing a risk assessment liable to ac- have a greater ratio of pigs per worker with less atten- count for other potentially harmful pathogens [25]. tion paid to individual finisher pigs, as suggested by Given the limitations stated above, we used a different Agostini et al. [22]. Some of the workers in bigger farms approach by grouping the farms according to their bio- were many times dedicated to jobs that are externalized security practices and then comparing their productive in smaller farms like general maintenance or feed manu- performance instead of directly modelling performance. facturing. This area needs further research to specify the In a multivariate approach to the data, farms were sep- types of staff in pig farms and its effects on health and arated in three clear clusters based on their biosecurity performance. The size of the farm may also have an ef- practices. The highest production performance was fect independent of the number of workers. Gardner, found in farms from clusters with better external biose- Willeberg and Mousing [23] described the duality faced curity but not internal biosecurity. External biosecurity by bigger farms which, facing higher risks of infection practices are easier to implement (fences, barriers, etc) due to frequent animal movements and high pressure of than internal biosecurity practices (i.e. attitudes and be- infection but having higher biosecurity standards to haviours). The former hint investment and could be minimize those risks. Finally, better disease manage- linked to better management and the maintenance of ment, including herd health protocols and veterinary health status, which translate into better performance. expertise, were linked to decreased mortality. The corre- Internal biosecurity practices, as discussed before, are lations between finisher mortality and areas with lower probably under estimated or not well understood by pig density and management of the nursery unit seem workers who lack a basic understanding of infectious contradictory and cannot be explained although no con- diseases. founding effects were found. As for the ADG model, it explained 16% of the vari- ability. Average herd size and experience of farm man- Limitations of the study agers had a negative impact on ADG, and a better Although this study accounts for almost 30% of all the disease management was positively correlated to ADG. breeding sows in Ireland, these herds were likely to rep- The negative impact of herd size in growth rate could be resent a better end of the Irish pig farms, as suggested related to the association with finisher mortality. In by Staaveren et al. [26]. Also, the biosecurity data was herds with higher disease pressure, growth rates are de- collected in a cross-sectional study in in-office inter- creased [24]. Other factors such as herd health or vac- views which may have led to bias towards answers stat- cination protocols may have a role in this association. ing measures believed to be applied on farm rather than The negative impact of experience could be related to stating measures applied [18]. Contributing to this bias several factors. Laanen et al. [5] found that older farmers was also the different interviewees with farm owners be- were associated with older infrastructures and poor in- ing less likely to be aware of the daily management prac- ternal biosecurity which could result in a lesser ability to tices and actual cleaning routines in their farms when address production challenges. compared to farm managers and other workers. The use The associations found between biosecurity categories of the Biocheck.UGent™ tool allowed an easy bench- and productive performance suggests that, in general, marking between countries; however it also avoided cap- farms with good biosecurity had better performance. turing some variation specific to each national context Laanen et al. [5] identified such associations with ADG and may be missing details of important internal biose- and FCR, but not with finisher mortality. Further simi- curity practices like replacement management or piglet larities between that study and ours are the low R2, management. Finally, as this was an observational study, meaning only a small proportion of the variability of the causal relationships should not be inferred from the re- productive performance was explained by biosecurity sults presented. Rodrigues da Costa et al. Porcine Health Management (2019) 5:4 Page 8 of 9 Conclusions Author details 1 This study assessed biosecurity practices in Irish pig Pig Development Department, Teagasc - Animal & Grassland Research and Innovation Centre, Moorepark, Fermoy, Co. Cork, Ireland. 2Departament de farms and its impact on production performance. Irish Ciència Animal i dels Aliments, Facultat de Veterinaria, Universitat Autònoma farms follow European tendencies with internal biose- de Barcelona, Bellaterra, 08193 Barcelona, Spain. 3Department of curity posing as the biggest liability. Our results sug- Reproduction, Obstetrics and Herd Health, Veterinary Epidemiology Unit, Faculty of Veterinary Medicine, Ghent University, Salisburylaan 133, 9820 gested that practices related to the environment and Merelbeke, Belgium. 4Pig Development Department, Teagasc Oak Park, Co. region, feed, water, and equipment supply and the man- Carlow, Ireland. 5School of Veterinary Medicine, University College Dublin, agement of the different stages, need to be particularly Belfield, Dublin 4, Ireland. addressed in poor performing farms to improve product- Received: 3 August 2018 Accepted: 21 December 2018 ivity of the Irish pig sector. Indeed such recommenda- tions should be adapted to the health status of each farm. Further studies on the economic impact of these References biosecurity practices in connection to data on herd 1. DAFM. Report of the pig industry stakeholder group. 2016. https://www. agriculture.gov.ie/media/migration/farmingsectors/pigs/ health, vaccination and feeding strategies are the key to REPORTPIGINDUSTSTAKEHOLDERGROUP290116.pdf. Accessed 3 July 2018. motivate farmers to change their practices. 2. Central Statistics Office. Meat Supply Balance. https://www.cso.ie/en/ releasesandpublications/er/msb/meatsupplybalance2016/. Accessed 17 Abbreviations June 2018. ADG: Average Daily Gain (g/day); d: Day; ePM: Teagasc e-ProfitMonitor; 3. Postma M, Backhans A, Collineau L, Loesken S, Sjolund M, Belloc C, FCR: Feed Conversion Ratio; g: grams; kg: kilograms; No.: Number; et al. The biosecurity status and its associations with production and PCA: Principal components analysis management characteristics in farrow-to-finish pig herds. Animal. 2016; 10:478–89. 4. Rojo-Gimeno C, Postma M, Dewulf J, Hogeveen H, Lauwers L, Wauters E. Acknowledgements Farm-economic analysis of reducing antimicrobial use whilst adopting The authors would like to thank Pilar Guzman Medina for her help with data improved management strategies on farrow-to-finish pig farms. Prev Vet collection, and the Teagasc advisors for facilitating the contact with farmers. Med. 2016;129:74–87. We would also like to thank all the farmers who accepted to take part in this 5. Laanen M, Persoons D, Ribbens S, de Jong E, Callens B, Strubbe M, et al. study. Relationship between biosecurity and production/antimicrobial treatment characteristics in pig herds. Vet J. 2013;198:508–12. Funding 6. Postma M, Vanderhaeghen W, Sarrazin S, Maes D, Dewulf J. Reducing This study was part of the PathSurvPig project which was funded by the antimicrobial usage in pig production without jeopardizing production Department of Agriculture Food and the Marine under the Research parameters. Zoonoses Public Health. 2017;64:63–74. Stimulus Fund (PathSurvPig 14/S/832) as part of the National Development 7. Amass SF, Clark LK. Biosecurity considerations for pork production units. Plan. JACD’s position was funded by the same project. The funders had no Swine Health Product. 1999;7:217–28. role in study design, data collection and analysis, decision to publish, or 8. Julio Pinto C, Santiago Urcelay V. Biosecurity practices on intensive pig preparation of the manuscript. MRC was supported by the Teagasc Walsh production systems in Chile. Prev Vet Med. 2003;59:139–45. Fellowship Fund. 9. Biocheck.UGent™, prevention is better than cure. www.biocheck.ugent.be. Accessed 20 Mar 2018. Availability of data and materials 10. Filippitzi ME, Brinch Kruse A, Postma M, Sarrazin S, Maes D, Alban L, et al. The datasets used and/or analysed during the current study are available Review of transmission routes of 24 infectious diseases preventable by from the corresponding author on reasonable request. biosecurity measures and comparison of the implementation of these measures in pig herds in six European countries. Transbound Emerg Dis. Authors’ contributions 2017;65:381–98. MRC: data collection, curation, statistical analysis and interpretation; manuscript 11. Kruse AB, Nielsen LR, Alban L. Herd typologies based on multivariate writing and reviewing. JGG: manuscript writing and reviewing. JACD: data analysis of biosecurity, productivity, antimicrobial and vaccine use data from collection and manuscript reviewing. MP and JDW: manuscript reviewing. GM: Danish sow herds. Prev Vet Med. 2018. data collection and manuscript reviewing. EGM: study design, funding 12. Backhans A, Sjölund M, Lindberg A, Emanuelson U. Biosecurity level and acquisition, data collection, statistical analysis and interpretation, manuscript health management practices in 60 Swedish farrow-to-finish herds. Acta Vet writing, reviewing and editing. All authors read and approved the final Scand. 2015;57:1–11. manuscript. 13. Teagasc. National Pig Herd Performance Report 2016. 2017. https://www. teagasc.ie/publications/2017/national-pig-herd-performance-report-2016. php. Accessed 4 July 2018. Ethics approval and consent to participate 14. AHDB. 2016 Pig cost of production in selected countries (InterPIG). 2017. All the farmers participating in this study gave individual signed consent to the https://pork.ahdb.org.uk/media/274535/2016-pig-cost-of-production-in- use of the data collected on farm, and to the retrieval of their production data selected-countries.pdf. Accessed 3 July 2018. from the Teagasc e-ProfitMonitor according to Teagasc’s internal data 15. R Core Team. R: A language and environment for statistical computing. 3.4.4 protection regulation. (2018-03-15). https://www.R-project.org/. Accessed 15 Mar 2018. 16. Hebbali A. olsrr: Tools for teaching and learning OLS regression. R package Consent for publication version 05.0. https://CRAN.R-project.org/package=olsrr. Accessed 15 June Not applicable. 2018. 17. Harrel Jr FE. rms: Regression Modeling Strategies. R package version 5.1–2. Competing interests https://CRAN.R-project.org/package=rms. Accessed 15 June 2018. The authors declare that they have no competing interests. 18. Casal J, De Manuel A, Mateu E, Martin M. Biosecurity measures on swine farms in Spain: perceptions by farmers and their relationship to current on- farm measures. Prev Vet Med. 2007;82:138–50. Publisher’s Note 19. Boklund A, Alban L, Mortensen S, Houe H. Biosecurity in 116 Danish Springer Nature remains neutral with regard to jurisdictional claims in fattening swineherds: descriptive results and factor analysis. Prev Vet published maps and institutional affiliations. Med. 2004;66:49–62. Rodrigues da Costa et al. Porcine Health Management (2019) 5:4 Page 9 of 9 20. Muns R, Nuntapaitoon M, Tummaruk P. Non-infectious causes of pre- weaning mortality in piglets. Livest Sci. 2016;184:46–57. 21. Kirkden RD, Broom DM, Andersen IL. Piglet mortality: management solutions. J Anim Sci. 2013;91:3361–89. 22. Agostini PS, Fahey AG, Manzanilla EG, O’Doherty JV, de Blas C, Gasa J. Management factors affecting mortality, feed intake and feed conversion ratio of grow-finishing pigs. Animal. 2014;8:1312–8. 23. Gardner IA, Willeberg P, Mousing J. Empirical and theoretical evidence for herd size as a risk factor for swine diseases. Anim Health Res Rev. 2002;3:43–55. 24. Cornelison AS, Karriker LA, Williams NH, Haberl BJ, Stalder KJ, Schulz LL, et al. Impact of health challenges on pig growth performance, carcass characteristics, and net returns under commercial conditions. Translat An Sci. 2018;2:50–61. 25. Iowa State University. PADRAP - Production Animal Disease Risk Assessment Program. https://vetmed.iastate.edu/vdpam/about/production-animal- medicine/swine/padrap. Accessed 3 July 2018. 26. van Staaveren N, Teixeira DL, Hanlon A, Boyle LA. Pig carcass tail lesions: the influence of record keeping through an advisory service and the relationship with farm performance parameters. Animal. 2017;11:140–6.