Herd-level risk factors for chronic pleuritis in finishing pigs: a case-control study

Background Chronic pleuritis is a common finding in slaughtered pigs in post-mortem meat inspection. The prevalence of pleuritis has been increasing during the last decade also in Finland. The aim of this prospective case-control study was to search for environmental, infectious and management-related herd-level risk factors for pleuritis in the slaughterhouse. Altogether 46 Finnish pig herds, including 25 control (low pleuritis prevalence in meat inspection) and 21 case (high pleuritis) herds, were enrolled in the study and visited during the tenth week of the rearing period of finishing pigs. Herd personnel were asked about basic herd information, management and environmental factors. Selected pigs were examined clinically, environmental parameters were measured and 15 blood samples per herd were taken during herd visits. Antibodies against Actinobacillus pleuropneumonia serotype 2 (APP2) and ApxIV toxin and swine influenza virus were measured. After the slaughter of study pigs, meat inspection results of the batch were gathered from slaughterhouses. Multivariate logistic regression models were built to identify possible risk factors for a herd to be a case herd ( i.e. having high pleuritis values). Results Finishing herd type, herd size and APP2 seropositivity were observed to act as risk factors. In addition to these, general herd-level factors, flank biting and high APP2 antibody prevalence of the herd tended to be associated with the risk of the herd being a case herd. None of the other clinical signs of pigs, management-related factors or environmental measurements were associated with herd status.


Abstract
Background Chronic pleuritis is a common finding in slaughtered pigs in post-mortem meat inspection. The prevalence of pleuritis has been increasing during the last decade also in Finland. The aim of this prospective case-control study was to search for environmental, infectious and management-related herd-level risk factors for pleuritis in the slaughterhouse. Altogether 46 Finnish pig herds, including 25 control (low pleuritis prevalence in meat inspection) and 21 case (high pleuritis) herds, were enrolled in the study and visited during the tenth week of the rearing period of finishing pigs. Herd personnel were asked about basic herd information, management and environmental factors. Selected pigs were examined clinically, environmental parameters were measured and 15 blood samples per herd were taken during herd visits. Antibodies against Actinobacillus pleuropneumonia serotype 2 (APP2) and ApxIV toxin and swine influenza virus were measured. After the slaughter of study pigs, meat inspection results of the batch were gathered from slaughterhouses. Multivariate logistic regression models were built to identify possible risk factors for a herd to be a case herd ( i.e. having high pleuritis values).
Results Finishing herd type, herd size and APP2 seropositivity were observed to act as risk factors. In addition to these, general herd-level factors, flank biting and high APP2 antibody prevalence of the herd tended to be associated with the risk of the herd being a case herd. None of the other clinical signs of pigs, management-related factors or environmental measurements were associated with herd status.
Conclusions As previously known, in endemic and subclinical infections such as APP, herd and management-related factors are important in building up infection pressure, but single risk factors seem to be difficult to identify. However, as flank biting was more common in high pleuritis herds, part of disease susceptibility is likely mediated via stress.

Background
Chronic pleuritis is a common finding in slaughtered pigs in post-mortem meat inspection. Pleuritis prevalence around 20% has been detected in several European countries [1,2,3,4]. In Finland, pleuritis prevalence in slaughter pigs has clearly been increasing during the last decade (see Fig. 1).
The registration of pleuritis has continuously increased also in a neighbouring country of Finland, Sweden, since the year 2000 [5].
Even though respiratory disease complex is caused by various pathogens, especially dorsocaudal pleuritis has been reported to be strongly associated with Actinobacillus pleuropneumoniae (APP) [1,6]. Also, swine influenza virus (SIV) has been noted to have an association with pleurisy [7]. The type and severity of the disease caused by APP are considered to depend on factors related to pathogen virulence, host, environment and management. Jirawattanapong et al. [9] found no single cause of pleuritis in their study in the Netherlands investigating possible infectious causes. They suggested the combined cause to be a variety of infectious agents together with environmental factors. While experimental studies have helped us to gather information about the pathogens, knowledge of the interplay of different factors affecting disease outcome in commercial pig farms remains incomplete [8]. Hence, the swine producing community lacks practical and effective means for pleuritis control.
The aim of this epidemiological case-control study in commercial swine herds was to search for various environmental, management-related and infectious factors increasing risk for high pleuritis in slaughtered pigs. If the main risk factors at herd level were known, we could target control measures more effectively.

Methods
This study was a prospective case-control study with a herd as the unit of interest. The target population included medium-to large-sized (more than 500 pigs per herd) Finnish herds rearing finishing pigs.

Sampling and data gathering
Purposive sampling was used. A study herd needed to fulfill the inclusion criteria of 1) at least 1000 finishing pigs sent for slaughter annually and 2) location in south-western Finland within a distance of 250 km from the University of Helsinki ambulatory clinic in Mäntsälä. Three major slaughterhouses were asked to compile a list of finishing pig herds fulfilling the above criteria. This list including 219 herds served as a sampling frame. At the same time, the slaughterhouses provided the research group with the pleuritis percentage of these herds during the year preceding enrolment in the study.
The herds were then sorted in descending order according to their pleuritis percentage, separately for each slaughterhouse. Herds were tentatively divided into case and control herds based on their pleuritis values. Herds having a higher pleuritis value than slaughterhouse-specific mean pleuritis plus standard deviation value were considered to be tentative case herds. Similarly, herds having a lower pleuritis value than the slaughterhouse mean minus the standard deviation were considered to be tentative control herds.
Researchers then contacted the tentative case and control herds and requested their consent to participate in the study. During the first contact researchers ascertained that the herd had not suffered from acute respiratory disease outbreak during the last batch of finishing pigs. All voluntary herds were enrolled in the study until the desired number of herds (at least 60 herds proportionally representing all three slaughterhouses) was fulfilled. Of 93 herds contacted, 29 (31.2%) opted out or were unable to participate. Out of herds that opted out, 48% were not willing to participate, 48% no longer had animals or had changed slaughterhouse, making participation impossible, and 3% reported some other reason. Customers of slaughterhouse 1 and control herds (54% controls vs. 45% cases) were overrepresented in herds opting out. Opted-out herds had approximately twice as many animals as herds that were willing to participate.
Researchers visited all study herds once about ten weeks after the growers (body weight 25 kg) had arrived at the finishing pig herd or a finisher compartment in a farrow-to-finish herd. At the beginning of the visit, the researchers ascertained that the pig groups had not suffered from acute respiratory disease outbreak after entering the finishing herd or compartment. The herd owner or responsible caretaker was interviewed about basic information of the herd (herd type, number of animals), management and environmental factors using a herd visit questionnaire. The main categories of management-related information included vaccination routines, biosecurity and hygiene, animal flow, mean number of animals per compartment and per pen, mean size of pen, medication routines and feeding. The main categories of environmental factors were ventilation, temperature, heating and floor type.
In each herd, one finishing pig compartment containing at least 200 but less than 1000 finishing pigs was chosen for clinical inspection of pigs and environmental measurements. After carefully entering the compartment, the researchers counted the number of pens containing pigs lying on top of each other. After that, pen wall temperature (Testo 830-T1, Testo SE  Altogether 15 pigs were selected evenly in different parts of the study compartment for blood sampling. These pigs were caught with a snout snare and a blood sample was taken from their vena jugularis with vacuum needles and serum tubes. The samples were centrifuged the next day in the laboratory, and the serum was stored at -20 °C until analysed.
At the end of the visit, outdoor temperature was measured. Before leaving the herd, herd personnel were asked to inform researchers about any major unexpected events, e.g. acute disease outbreak or equipment malfunction, during the remaining rearing period.
After the batch of finishing pigs was sent to slaughter, the slaughterhouses provided the researchers with the meat inspection (MI) findings of each batch, including mean carcass weight, kilograms of condemned meat, percentage of whole and partial carcass condemnations, lean meat percentage, abscesses, arthritis, milk spots in liver, organ condemnations, tail biting, pneumonia and pleuritis.
Herds having pleuritis percentage higher than the slaughterhouse-specific mean pleuritis plus one standard deviation in the batch undergoing analysis were considered batch-specific case herds.
Similarly, herds having pleuritis percentage lower than the slaughterhouse mean minus one standard deviation were considered batch-specific control herds. Finally, the group of confirmed case herds included only tentative case herds that were also batch-specific case herds. Similarly, confirmed control herds included only those tentative control herds that were also batch-specific control herds.
Laboratory analysis APP antibodies were measured using two commercial test kits: IDEXX APP-ApxIV ELISA (IDEXX, Liebefeld-Bern, Swizerland) to detect antibodies against ApxIV toxin, which is produced by all known APP serotypes (19), and IDvet ID Screen APP 2 indirect ELISA (IDvet, Grabels, France) to detect antibodies against lipopolysaccharides (LPS) specific to APP serotype 2 (APP2), with a sensitivity of 82.9% and a specificity of 99.6% for IDEXX APP ApxIV ELISA and a specificity of 99.68% for IDVet APP2 ELISA. A pig was considered positive if the test used detected any antibodies in the serum sample.
All blood samples were tested with influenza A antibody ELISA (ID Screen® Influenza A Antibody Competition, IdVet, Grabels, France) according to the manufacturer's instructions. A sample was considered unclear when the competition percentage (S/N%) was 46-49% and positive when the competition percentage was ≤ 45%. If a herd had at least one unclear or positive blood sample (pig) in the ELISA test, blood samples of that herd were further analysed using a haemagglutination inhibition (HI) test according to the European Surveillance Network for Influenza in Pigs. This was done with the antigens H1N1 (SW/Best/96), H1N2 (SW/Gent/7625/99) and H3N2 (SW/St. Oedenrode/96). All antigens were provided by GD Animal Health Service (Deventer, the Netherlands).
A sample was considered HI positive if the HI titre was ≥ 1:20.
Further details of serology can be found in the article by Haimi-Hakala et al. [10].

Statistical analysis
A required sample size of 24 herds in both groups (control and case) was calculated assuming the proportion of exposed as 40% for controls and as 80% for cases in presumably the most influential variable (herd type). Alpha 0.05 and power 0.8 were used. The least extreme odds ratio to be detected is 6.0 [11].
All gathered data were scrutinized, and all unreliable answers were either checked and corrected, or if this was impossible, removed from the dataset. Most of the variables describing management-related and environmental factors were transformed into meaningful categories. Most of the count variables and measurements were handled as continuous variables.
The outcome variable was a categorical variable "confirmed case or control". Descriptive statistics of all predictor variables (management-related factors, clinical signs, environmental measurements) were compiled containing all herds, and a comparison between case and control herds was carried out.
For the analysis of management-related factors, univariate associations between predictor variables and outcome were evaluated using logistic regression. The liberal p-value of 0.2 was used as a keepin or drop-out threshold. The correlations between predictor values were scrutinized. The herd type (farrow-to-finish or finishing only) was detected to correlate strongly with many predictor variables (e.g. all-in all-out production). The decision to keep the variable "herd type" in further analysis and drop correlating, intervening variables was made.
Finally, a multivariable logistic regression model for herd-level management-related data gathered by the interview was built (model 1). The initial model 1 contained predictor variables: number of finishing pigs in the herd (< 1000/≥1000 pigs), compartment disinfection (always between batches, sometimes, never), littering frequency (once or twice per day or continuously available), proportion of slatted flooring (≤ 50%/>50%), airspace per pig, feeding type (liquid/dry), piggery temperature when weaners entered the compartment, piggery temperature when finishing pigs left the compartment, heating (yes/no), ventilation system service (yes/no), ventilation adjustment difference in winter, loading corridor (yes/no) and handwashing facility for visitors (yes/no). Backward elimination model building strategy was then utilized. The final model 1 contained only the predictor variables herd type and number of finishing pigs per herd.
An alternative way of handling the data was also tested (model 2). The variables related to hygiene, animal flow management and biosecurity were scrutinized subjectively, and three combined variables were formed: hygiene (good/poor), animal flow management (good/poor) and biosecurity (good/poor).
As the variables related to ventilation were troublesome to value as there is no relevant knowledge about practices used in piggery ventilation, those variables were kept separately, and no combined variables were formed. A multivariate model utilizing three combined variables, i.e. variables related to ventilation, number of finishing pigs and herd type, was built. Like model 1, only number of finishing pigs and herd type remained in model 2, and therefore, the results of model 2 are not presented.
For the analysis of environmental measurements, a multivariable logistic model was built (model 3).
All other variables were modelled as continuous, but variable NH3 concentration was categorized.
None of the variables were revealed to be statistically significant. For modelling serological results for APP, herd-level APP2 prevalence and APPIV antibody prevalence were calculated. A logistic regression model was built for both serological tests (APP2 and APPIV) separately. The final models 5 and 6 for APP contained prevalence (%) of seropositive pigs in a herd as a categorical variable (low/high prevalence within herd, mean used as a cut-off point), herd type as a confounder and number of finishing pigs as a categorical variable (≤ 1000/>1000 pigs). For SIV serology, univariate association with herd pleuritis status was clearly insignificant, and therefore, no further modelling was performed.
For brief model diagnostics, the basic assumptions of logistic models were inspected with regard to data structure and nature of the outcome variables. In addition, residuals were scrutinized. No serious breaches of the underlying assumptions were detected. Furthermore, the area under the ROC curve was evaluated to assess predictive ability of models. For models revealing significant results, the values ranged from 0.6 to 0.9.

Results
Originally, data were available from 64 herds (28 tentative case and 36 tentative control herds). We removed 18 of the herds because they did not have the same pleuritis status in the batch inspected after the herd visit. Therefore, 46 study herds remained, 25 control and 21 case herds, for data analysis. These herds were belonged in three different slaughterhouses (slaughterhouse 1, 27 herds, 59%; slaughterhouse 2, 14 herds, 30%; slaughterhouse 3, 5 herds, 11%). A slight majority of herds reared only finishing pigs (26 herds, 57%) and the rest were farrow-to-finish herds (20 herds, 43%).
Basic descriptive statistics of study herds (n = 46) are shown in Table 1. All-in all-out production (AIAO) was utilized in 24 herds (52%) and was not utilized in 22 herds (48%). In control and case herds, AIAO was used in nine (36%) and 15 (71%) herds, respectively. The AIAO production was associated significantly with pleuritis in univariate analysis (p = 0.02).  The summary of environmental measures in study herds has been collected in Table 3 for the continuous variables. The only categorical environmental variable was NH 3 level; the ammonia concentration was < 5 ppm in 20 herds (43.5%) and 5-18 ppm in 26 herds (56.5%). In control herds, the ammonia concentrations were < 5 ppm and 5-9 ppm in 12 (48.0%) and 13 (52.0%) herds, respectively. In case herds, the concentration was < 5 ppm in 8 (38.1%) and 5-18 ppm in 13 herds (61.9%). The summary of meat inspection findings in the study herds after the herd visits is presented in Table 4. The prevalence of pleuritis with the slaughterhouse and herd classification is presented in Table 5.  SIV serology was carried out in 40 of 46 study herds for 600 blood samples. Five herds (12.5%), specifically one control and four case herds, had at least one pig with antibodies against SIV.

Results from statistical modelling
Regarding possible management-related risk factors for a herd being a case herd (i.e. high pleuritis prevalence at slaughter), the herd type "farrow-to-finish" was revealed to be a protective factor compared with the herd type "only finishing pigs". Furthermore, a herd having more than 1000 fattening pigs (compared with less than 1000 fattening pigs) had greater odds of being a case herd.
Results of the logistic regression model of management-related risk factors (model 1) for pleuritis are presented in Table 6. Legend: The reference level for categorical variables is provided in parentheses.
Regarding environmental measurements, statistical modelling did not reveal any associations between measurements and herd status (model 3).
A herd having higher than mean study herd prevalence of flank biting among study animals (compared to less than mean study herd prevalence of flank biting) tended to have higher odds of being a case herd. Results of logistic regression model of clinical signs (model 4) in the rearing week are presented in Table 7. Legend: The reference level for categorical variables is provided in parentheses.
High prevalence of APP2 antibodies tended to be associated with case herds (OR 7.8, p = 0.1, model 5). No association was found regarding APPIV antibody and herd status (model 6).

Discussion
The study revealed some important factors to be associated with high pleuritis values in pig herds in a country with a low prevalence of many respiratory diseases. A protective effect of "farrow-to-finish" herd type compared with "only finishing pigs" herds emerged regarding odds of having higher pleuritis values in meat inspection. In previous studies, contradictory results have been obtained when herd type has been considered in association with pleuritis. Several studies have found that herds that purchase weaners (i.e. finishing pig herds) have a greater risk for respiratory diseases than herds with closed production [2,12]. On the other hand, farrow-to-finish herds have been observed to have greater odds for having chronic pleuritis in slaughtered pigs [13,14]. A fairly recent study by Jäger et al. [13] found that if growers are purchased from more than three different farms the protective effect of finishing herd over farrow-to-finish herd was diminished. In addition to herd type, the AIAO production system has been associated strongly with better respiratory health and, more specifically, lower prevalence of pleuritis [13,15]. In our study, AIAO production correlated strongly with herd type, and because of this the variable was omitted from multivariate modelling. It is not at all surprising that finishing pig herds were able to empty the piggery at one time and farrow-to-finish herds utilized more continuous flow of animals. In our study, AIAO production was significantly associated with pleuritis in univariate analysis, but surprisingly, case herds utilized more often the AIAO system than control herds. When we investigated only the effect of AIAO production on level of pleuritis and built a statistical model solely for this purpose, our data (results not shown) revealed that herd type acts as confounding variable. When confounding is taken into account, AIAO production seems to have a protective (albeit not significant) effect on pleuritis level.
The present study agrees with the previously described debilitating effect of growing herd size on prevalence of pleuritis [5,6,7,14,16]. This effect could be related to infection pressure because bigger herds are more likely to need to purchase more animals, which is accompanied by an increased risk of introducing pathogens or naïve animals into the herd. Disease dynamics (i.e. better possibilities for spread and maintenance of airborne infection) also differ between bigger and smaller herds [12]. Considering environmental risks, it might also be more difficult for bigger herds to control optimal air quality, especially if the compartments are large. However, this was not the case in our study, where room size did not differ between large and small herds.
Herd type and size are commonly and inextricably associated with specific types of management practices, e.g. farrow-to-finish herds do not buy weaners or growers and finishing herds more commonly are able to practice AIAO production. However, the herd type itself, when used in the sense of sourcing animals, seems to be too vague a definition for choices made on a certain farm regarding infection pressure. To be able to reliably predict risk factors for chronic pleuritis, these management practices need to be defined in more detail. Earlier studies have been able to find several risk factors for pleuritis in the slaughterhouse such as number of pigs per pen [2], pig density in the neighbourhood [3,4], low health status of the herd [3], poor biosecurity [4], lack of disinfection of the farrowing room [6], no cleaning and disinfection [13], lack of complete AIAO production [3,13,17], mixing of pigs [3,13], season [3,4,18], mean temperature below 23 °C in the finishing unit [6], low weaning age [2] and low airspace [4]. Jäger et al. [13] observed that keeping pigs with more than a one-month age difference in the same airspace acted as a risk factor for pleuritis. In our study, we observed a marked weight difference (approximately 40-50 kg) among finishing pigs kept in the same room, which had an association with pleuritis in univariate analysis but did not remain in the multivariate model. However, the weight difference was more marked in control than case herds. The weight difference registered was a visual approximation between the smallest and the biggest pig in the room, and no actual weighing of pigs was used. This kind of method might have accentuated extreme observations and might contain considerable error. Cleveland-Nielsen et al. [3] found that feeding only dry feed protected the herd from high pleuritis values. However, some studies have not observed any association between non-infectious risk factors studied and pleuritis in the slaughterhouse [1,19]. Some of the significant associations between pleuritis and management or health factors have been found only in univariate analyses, while in multivariable analysis the associations have disappeared. This highlights the need to be able to model several correlated variables simultaneously. Hurnik et al. [15,20] attempted to overcome these kinds of problems typical for surveys by using a factor analysis. Regarding pleuritis, they found only one common observed feature of herds having greater prevalence of pleuritis: extensive type of farming. However, there is convincing evidence in the literature, as summarized above, that management choices related to infection pressure have a considerable effect on prevalence of chronic pleuritis. This is further supported by the observation of the protective effect of farrow-to-finish herd type and small herd size in our study.
Flank biting was more prevalent in case herds than in control herds in the present study (2.6% vs. 0.04%); moreover, in the regression model flank biting tended to be associated with higher pleuritis prevalence of the herd (OR 9.6, p = 0.05). No previous studies have reported an association between flank biting and meat inspection findings in slaughtered pigs. However, some studies have recognized the association between tail biting, another vice common in pigs, and prevalence of pleuritis. Tail biting seems to act as a risk factor for pleuritis found in meat inspection [21,22,23]. The prevalence of both acute and healed tail biting was higher in our case herds than in control herds. However, the difference was not statistically significant, probably because of marked variation between herds and a lack of statistical power. The association between behavioural vices and disease susceptibility is thought to be mediated through stress, which in general is considered to affect immunity negatively, thus making the host animal more susceptible to diseases [24]. Recent evidence suggests that environmental factors, namely enriched housing, probably via decreased stress, can reduce susceptibility to, for instance, co-infection with porcine reproductive and respiratory virus (PRRRSv) and APP [25]. Furthermore, it is unlikely that a respiratory pathogen, such as APP, is transmitted by flank or tail biting. It is more plausible that flank or tail biting share similar risk factors with pleuritis, e.g. overcrowding, lack of straw, large group size, climatic extremes, feeding type and poor ventilation [26]. In addition, Klinkerberg et al. [27] and Tobias et al. [28] have carried out simulation and transmission studies and have concluded that APP outbreaks are unlikely to be caused by spread of the pathogen, but more likely to be a consequence of clinical signs triggered in pigs already infected. The observed tendency in our study that links flank biting with chronic pleuritis supports the importance of management factors in avoiding stress in commercial pig farms, in this way decreasing susceptibility to respiratory disease.
Most studies investigating pleuritis have been done in countries with many possible pathogens present in the pig population. In Finland, the prevalence of porcine respiratory pathogens differs from the situation in continental Europe. The country has been free from Aujeszky disease virus (ADV), porcine respiratory corona virus (PRCV) and porcine reproductive and respiratory syndrome virus (PRRSV) for decades and nearly free from Mycoplasma hyopneumoniae (MHyo) [10]. Porcine respiratory disease complex is a multifactorial syndrome with clinical signs caused usually by multiple micro-organisms, both bacteria and viruses together with environmental and management-related factors as well as genetics [29]. Both viruses and MHyo have been considered as primary pathogens, which predispose pigs to concomitant bacterial infections such as APP [29]. Earlier, seropositivity to MHyo was reported to increase the odds of contracting chronic pleuritis [14], and APP serotype 2 together with PRRSV was described to be significantly associated with pleuritis [30].
We found a tendency for high prevalence of APP2 antibodies in a herd to be associated with higher prevalence of pleuritis (OR 7.8, p = 0.1). However, this was not the case for APPIV antibodies.
Controversial results have been previously reported regarding serology and pleuritis registered in meat inspection [2,9,14,31]. In case of endemic disease, serology might not be the most useful tool in diagnosis or at least paired samples are needed [10]. Although we observed high pleuritis values and high prevalence of APP antibodies, APP might not be the main causative infectious agent in our study herds. Also, the timing of the infection remains unknown in our study, as we are dealing with endemic disease. For example, Wallgren et al. [5] found different serological patterns during the growing time of pigs in four herds. They showed that repetitive sampling helped to pinpoint the actual causative agent and disease pattern in different herds. In addition, APP serotypes vary not only in commonness, but also in virulence. In Finland, the virulent serotype 2 has been found to be common throughout the country [32], causing also acute respiratory infections (10). Fablet et al. [30] noted that of the APP serotypes only serotype 2 was associated with pleuritis. Hence, it seems logical that antibodies, especially against serotype 2, tend to be associated with high pleuritis values. APxV toxin is produced by all APP serotypes and this might explain why the association of APPIV antibodies with pleuritis went undetected.
In our study, only five herds had antibodies against SIV. Even though four out of these five herds were case herds and only one a control herd, the total number of SIV-positive herds was too small for statistical modelling. SIV, both H1N1 and A(H1N1)pdm09, was first found in Finland in 2008 and 2009 [33], and both of the strains have thereafter spread throughout the country. The arrival of this new pathogen to a previously naïve population may have played a role in the increased pleuritis prevalence observed in meat inspection during recent years. However, based on current results, it remains unclear whether SIV plays a role in chronic pleuritis in Finland.
Different composition of primary respiratory pathogens may have an impact on the interplay of different risk factors and secondary pathogens such as APP. Without major primary pathogens, finishing pigs might be able to withstand some deficiencies in management without acute or chronic illness. Thus, in an observational study in a naïve country like Finland, the effect of risk factors might need to be more prominent than in other countries to cause detectable alterations in outcome.
In this study, herds were defined as cases or controls based on detection of pleuritis for a longer time period than only one batch. Furthermore, we gathered management-related data during herd visits, which enabled us also to inspect the pigs clinically. These choices should have made both the allocation of herds into cases or controls and the collection of data reliable. The results should be more valid than in studies where allocation has been done based on a single batch of slaughtered pigs or where the data have been collected with questionnaires sent to farmers or telephone interviews [3,13,17]. Purposive sampling was utilized in our study to highlight the differences between case and control herds. Typically, this type of sampling leads to valid estimates if otherwise conducted properly, but might restrict extrapolation of study results. Study herds had on average 1000 finishing pigs, which is approximately double the typical Finnish pig herd size. However, 50% of finishing pigs in Finland are produced in herds having more than 1000 finishing pigs. While descriptive statistics may not be representative of all Finnish pig herds, the results may well be applied in modern pig production. As always in a case-control study, the role of cause and consequence cannot be proven. However, most of the recorded factors were linked to reasonably permanent features of the study herds and present over a prolonged time period, and thus, could thought to precede the observed MI findings.
Almost 70% of herd owners asked to participate in the study were willing to be involved. As control herds were overrepresented amongst opt-out herds, there is a possibility of selection bias in the data.
The most frequent reason for opting out was "not willing to participate". Almost 60% of these herds were case herds and the reason for refusal was frequently that "participation will not help them to overcome the high pleuritis problem that they have". The majority of opt-out control herds no longer had pigs, and this was the second most common reason for opting out. Generally, opt-out herds had a larger herd size than opt-in herds. Therefore, it is possible that the herds not willing to participate in the study might have been not only larger, but also not as keen to take active measurements to improve their management. This kind of perceived attitude may have some (possibly negative) influence on overall management of herds. Possible selection bias may have caused study herds to resemble each other more in terms of influential management variables than they actually do in the source population, leading to an underestimation of the severity of risk factors.
Lack of statistical power might have caused several factors observed to differ between case and control herds to be seen as statistically non-significant. For example, for the present sample size (at least 24 herds in both groups) the presumed difference in exposed and non-exposed herds should have been 40% and 80% in most influential risk factor (herd type). The realized distribution (36% vs. 80%) fulfilled that criteria well and statistical significance was seen. However, as calculated for another categorical variable (NH3 concentration), the observed proportions of exposed and nonexposed herds were 61% and 52%, respectively, which may lead in a lack of statistical power.

Conclusions
Finishing herd type, herd size and APP2 seropositivity were observed to act as risk factors for a herd to be a case herd (i.e. a herd having high prevalence for chronic pleuritis detected in meat inspection). In addition to these general herd-level factors, flank biting and higher level of APP2 antibody prevalence of the herd tended to be associated with the risk of the herd being a case herd with higher prevalence of pleuritis. No other clinical signs of the pigs, management-related factors or environmental measurements were associated with pleuritis values As previously known, in cases of endemic and subclinical infections, such as APP, the herd and management-related factors are important in building up infection pressure, but single risk factors seem to be difficult to identify.
However, as the flank biting was more common in high pleuritis herds, part of the disease susceptibility is likely mediated via stress. Thus, shared risk factors for stress and lower respiratory health are worth further exploration utilizing modern analysis methods, such as factor analysis, capable of handling numerous and correlated variables.

Declarations
Ethics approval and consent to participate The experiment was approved by the southern Finland regional state administrative agency (ESAVI/8511/04.10.07/2013).

Consent for publication
Not applicable.

Availability of data and material
The datasets supporting the conclusions of this article are available in the Open Science Framework repository, https://osf.io/usmy7/.