Histomonosis in German turkey flocks: possible ways of pathogen introduction

ABSTRACT Histomonosis has become an important disease of turkeys since the ban of effective feed additives and therapeutics. Some critical risk factors for pathogen introduction into a farm have already been identified but open questions remain. Therefore, a retrospective case–control-study was used to identify the most significant risk factors for Histomonas (H.) meleagridis-introduction into a turkey farm. A total of 113 questionnaires were collected from 73 control-farms and 40 Histomonas-positive case-farms in Germany between 20 April 2021 and 31 January 2022. The data were analysed for possible risk factors by descriptive and univariate, single- and multi-factorial analysis. The presence of earthworms, snails and beetles, as vectors of H. meleagridis, as well as the proximity to other poultry-keeping farms in addition to a frequent observation of wild birds nearby the turkey farm, showed the highest risk potential for histomonosis outbreaks. Furthermore, poor biosecurity measures seem to have increased the probability for an outbreak. Insufficient climate management, straw as litter material and an inadequate litter refill frequency might have promoted a favourable humidity for vector- or pathogen survival providing important areas for improved disease control measures in the future. 
 RESEARCH HIGHLIGHTS
 A retrospective case–control-study was conducted to identify impactful risk factors for a H. meleagridis introduction. The probability of a histomonosis outbreak was increased by the presence of vectors and reservoirs nearby a farm. Impactful risk factors concerning biosecurity measures, climate and litter management were identified.


Introduction
Histomonosis has become a serious disease for turkeys as the use of effective feed additives and therapeutics has been banned for reasons of food safety in the European Union and other countries worldwide (Commission of European Communities, 1995, 2002, 2010Clark & Kimminau, 2017).
The disease-causing pathogen, Histomonas meleagridis (H. meleagridis), is a parasite that appears in two stages, non-amoeboid and amoeboid (Tyzzer, 1920;Cortes et al., 2004). A third, cyst-like stage was additionally observed in in vitro studies under unfavourable environmental conditions (Munsch, Hess et al., 2009;. The pathogen cannot survive on its own in the environment for a long time (Lotfi et al., 2012). However, it can reach increased survivability by inclusion in vectors (Lund et al., 1974). Besides the caecal worm Heterakis gallinarum and its contaminated eggs, earthworms serve as main transport hosts, in addition to other insects, such as flies and beetles, which can act as mechanical vectors (McDougald, 2005;Hauck et al., 2010;Hauck & Hafez, 2013). The pathogen can be ingested orally or via the cloacal route (Hauck & Hafez, 2013). After the caecal colonization of the pathogen and the subsequent inflammatory reactions, which can result in functional and structural damage of the gut barrier, a haematogenous dissemination to the liver and to other organs may occur (Powell et al., 2009;Hauck & Hafez, 2013).
The above-mentioned pathogenesis results clinically in a typical, sulphur-coloured diarrhoea as well as nonspecific signs, such as ruffled feathers and apathy (Senties-Cue et al., 2009;Windisch & Hess, 2010). In addition to turkeys, which often reach high mortality rates, other poultry species as well as wild birds can be infected with the pathogen and can develop clinical signs to a variable extent (McDougald, 2005;Hauck & Hafez, 2013).
There are a few studies on possible ways of pathogen introduction into a flock. Animate (e.g. earthworms, beetles and flies) and inanimate vectors (e.g. litter material, feed, soil and machine wheels) of H. meleagridis have already been identified (McDougald, 2005;Senties-Cue et al., 2009;Hauck et al., 2010;Hauck & Hafez, 2013). Other poultry-harbouring farms and wild birds could form a reservoir for H. meleagridis (Potts, 2009;Grafl et al., 2011;Popp et al., 2011;Jones et al., 2020;Noor et al., 2021). Preexisting diseases, as well as climate conditions, the change of feed, the sex and poor hygiene measures may be outbreak-favouring factors (Thompson, 1953;Callait-Cardinal et al., 2010;Clark, 2019;Sulejmanovic et al., 2019). However, in most of the studies, the abovementioned ways of pathogen introduction and outbreak-favouring conditions are only suspected to be histomonosis risk factors, as a confirmation cannot be provided in case and field reports (Aka et al., 2011).
Comprehensive and systematic investigations on possible ways of pathogen introduction and on outbreak-favouring conditions have not yet been carried out, although an increasing problem of histomonosis has been reported in many countries including Germany.
Therefore, the aim of our study was to systematically elucidate possible ways of H. meleagridis introduction into turkey farms and determine possible conditions which may favour an outbreak.

Study design and participants
A retrospective case-control-study was designed and conducted in Germany for the investigation of ways of pathogen introduction into a turkey farm and of outbreak-favouring conditions. As no complete list of all turkey farms in Germany was available, the study population was established by a convenience sampling. It was hypothesized that, in a 1:2-setting, the prevalence in the control-group was 50% and the expected odds ratio (OR) was on average 1.2. The sample size was calculated using PASS Power Analysis and Sample Size Software (NCSS, LLC. Kaysville, UT, USA, ncss.com/ software/pass). The confidence level (CI) was set to 95% with a two-sided-test strategy (method: Mantel-Haenszel). The calculated sample size was 423 (141 cases to 282 controls). As the target population was defined by German turkey farms with > 1000 turkeys, the calculated sample size represented 60% of the target population (Federal Statistical Office, 2021).
To collect data, a questionnaire was designed. Several guidelines and expert advice were integrated into the design of the questionnaire. After a successful testing phase, the questionnaire was published online via the platform LimeSurvey (LimeSurvey GmbH, Hamburg, Germany). Replies were collected between 20 April 2021 and 31 January 2022. Turkey farms throughout Germany, which were members of the association of German poultry industry (ZDG) and its sub-association of German turkey producers (VDP), were invited several times to answer the questionnaire.
No ethical approval was required, as the case-control-study was non-interventional.

Variables
The target population was represented by German turkey farms, which harbour more than 1000 turkeys (Supplemental Material 1). The farms were divided into two groups. Case-farms were defined as farms which had an outbreak of histomonosis since January 2018. The control-group consisted of farms without any outbreaks of histomonosis, and of farms which had a problem with histomonosis before January 2016. The classification of the groups was oriented at the time point of increased histomonosis field outbreaks in Germany, which was identified by personal communications with the VDP.
For analysis, data were exported to Microsoft Excel (Microsoft Corporation, Redmond, WA, USA, 2018. Microsoft Excel, https://office.microsoft.com/excel), checked for plausibility and imported into the statistical software. Fifty-four influencing factors were analysed. They were classified into five categories: turkey house conditions; vectors, reservoirs and related environmental conditions; biosecurity and hygiene; litter management; and farm management.

Confounding
In order to control for as many confounders as possible, the responses were collected in the ratio of about two control-farms to one case-farm. Possible confounders, which were still present, were integrated into the selection procedure of the final model (Dohoo et al., 2003).

Bias
The participation in the survey was performed on a voluntary basis. In addition to the online-participation, the questionnaire was sent by mail to 16 interested farmers, who did not have the opportunity to answer the online survey, to minimize selection bias.
The response categories were asked as precisely as possible to avoid misclassification and information bias. During the descriptive analysis, categories were merged meaningfully, based on their distribution, in order to fulfil the requirements of inferential statistics.
As the study design was retrospective, there is a high risk of recall bias. For this reason, we included only case-farms with an outbreak since January 2018.
A potential attrition bias was controlled by including the "No data"-category in the descriptive and inductive analyses.
To avoid publication bias, all collected data are represented in Tables 1-2 and Supplemental Material 1-3 of this paper. The most important results are discussed in more detail.

Statistical analysis
The statistical analysis was performed using SAS Enterprise Guide, version 7.15, and SAS Software, version 9.4 (SAS Institute Inc., Cary, NC, USA).
All variables were analysed descriptively (n and % for qualitative variables). Their possible relations were visualized in Figure 1 (created with Microsoft Power-Point, Microsoft Corporation, Redmond, WA, USA, 2018, https://office.microsoft.com/powerpoint). Unanswered questions were included as the category "No data" in all descriptive and inductive analyses. Fifteen variables were excluded from the inductive analysis after detailed consideration due to lack of variance (Supplemental Material 2). Inductive analysis was performed with 39 variables. The univariate, single-factorial analysis was performed using logistic regression (Table 1 and Supplemental Material 3). Thereafter, the variables were selected for a notable and plausible odds ratio (OR) (≤ 0.8 or ≥ 1.2), a plausible 95%-CI (between 0.0 and 20.0) and a P-value ≤ 0.2 (Dohoo et al., 2003; Table 1 and Supplemental Material 3, variables marked with "d"). Apart from bias checks, the results of the category "No data" were not interpreted with respect to their content. As a second selection procedure, the associations between the selected variables were tested using Fisher's exact test and the variationinflation-factor. The significance level for the Fisher's exact test was based at P ≤ 0.05 and for the variance inflation factor ≤ 10. The associations between the selected variables were significant in most of the cases (data not shown). Nevertheless, the contentual plausibility of associations could be proven in only one case. That is why one variable was excluded and the other selected variables were included in the univariate, multi-factorial model. Two possible confounders were also integrated before the selection procedure was done. A univariate, multi-factorial analysis was performed using logistic regression ( Table 2). The best model was determined by stepwise backward selection of the variables and confounders using the Akaike Information Criterion and by forward selection of the plausible interactions between variables using the model-fit criteria Akaike Information Criterion as well (Dohoo et al., 2003).

Study design and participants
One hundred and twenty-one farmers took part in the study by answering the questionnaire, but six of them only completed less than 50% of the questionnaire. These six questionnaires were evaluated separately. The answers were basically comparable to the remaining 115 participants. Furthermore, two as intermediate categorized farms (with a histomonosis outbreak between January 2016 and December 2017) answered the questionnaire. Due to the small group size, this group was excluded from further analyses. Therefore, the responses of 113 questionnaires, with 73 controlfarms and 40 case-farms, were included in the further analysis.
According to the newest report of the Federal Statistical Office, there were 701 turkey farms with more than 1000 turkeys in March 2020 in Germany (Federal Statistical Office, 2021). With a number of 113 out of 701 possible participants, the study had a general participation rate of about 16.1%. The participating farms were located in 10 federal states of Germany, with participation from the northern part as well as from the southern part of the country (Supplemental Material 1). The participation rate varied between 0.0% and 36.4% per federal state. Federal states with a low as well as a high density of turkey farms took part in this study.

Possible ways of pathogen introduction
Turkey house conditions Flocks kept in houses which were built after 1995 were 30% more likely to be affected by histomonosis than flocks kept in older houses (OR = 1.312, P = 0.702; Table 1).
A closed house-type doubled the probability for histomonosis on a farm as compared to houses of the Louisiana-type (OR = 2.080, P = 0.123).

Possible vectors, reservoirs and related environmental conditions
In general, a low frequency of insect control (45.0% of case-farms and 46.6% of control-farms; Supplemental Material 3) and snail control (27.5% of case-farms and 19.2% of control-farms; Table 1) was identified. However, observing earthworms as vectors or snails and beetles as potential mechanical vectors in the environment of the turkey house was associated with an increase in probability of becoming affected by histomonosis by 63.4% (P = 0.228) as compared to farms where no such vectors were observed.
Poultry flocks within a radius of 2 kilometres around the participating farms increased the probability of being affected by histomonosis by 2.362times (P = 0.079). The detailed survey showed that broiler flocks (OR = 1.215, P = 0.659), turkey flocks (OR not calculable, but notable descriptive distribution: 12.5% of case-farms and 0.0% of controlfarms with turkey flocks within a radius of 2 kilometres) and other types of poultry flocks (such as Table 1. Descriptive and univariate, single-factorial analysis of selected variables with a plausible and notable OR and CI; grouped according to possible ways of pathogen introduction and conditions, which favour an outbreak. quail, guinea fowl and backyard poultry; OR = 2.148, P = 0.116), increased the probability noticeably. The observation of songbirds, birds of prey and cranes nearby the participating farms increased the probability of an outbreak of histomonosis by 75.3% (P = 0.232), 22.5% (P = 0.828) and 304.6% (P = 0.101), respectively.
Farms with driveways made of paving stones were 54.4% more frequently affected by histomonosis than farms with concrete or asphalt driveways or both (P = 0.380). A similar result was found for gravel driveways, with a 2.844-times higher probability for a histomonosis outbreak (P = 0.035).
Emergency doors or maintenance doors or both leading to unfortified ground outside the turkey house without any hygiene barriers increased the probability of becoming affected by histomonosis by 71.4% compared to farms where no such doors were available (P = 0.316).
A humus-rich surrounding increased the probability of becoming affected by histomonosis by 21.1% as compared to farms with a sand-or loamrich surrounding or both (P = 0.661).

Biosecurity and hygiene
The usage of equipment outside the turkey house (OR = 0.462, P = 0.259) and consistent use of the same disinfectant in the house (OR = 0.480, P = 0.149) or no usage of disinfectants in the house (OR = 0.480, P = 0.305) reduced the probability of becoming affected by histomonosis by half.
Farms which cleaned or disinfected, or cleaned and disinfected, the wheels of their machinery only sometimes before entering the turkey house had a 1.404-higher probability of becoming affected by histomonosis than farms, which cleaned or disinfected, or clean and disinfected, the wheels of their machinery every time before entering the house (P = 0.453).
Farms without increased cleaning efforts after any case of illness were 66% more likely to become affected by histomonosis than farms with increased cleaning efforts (OR = 1.660, P = 0.294).

Litter management
In the univariate, single-factorial model (OR = 2.797, P = 0.015) and in the univariate, multi-factorial model (OR = 2.324, P = 0.057; Table 2) it was shown that the weather-dependent litter refill frequency doubled or nearly tripled the probability of an outbreak compared to farms with a fixed litter refill frequency.
Straw, as litter material, increased the probability of a histomonosis outbreak compared to other litter materials, referring to an OR of 2.656 (P = 0.101; Table 1).
Farms where litter material was stored outdoor or in an open hall or both had a 1.333-higher probability of becoming affected by histomonosis compared to farms where the litter material was stored in the turkey house itself or in the anteroom or both (P = 0.772).

Farm management
The probability of a histomonosis outbreak increased by 1.395-times (P = 0.442) if turkeys were fattened onsite starting at 4-6 weeks post hatch, compared to farms that started raising turkeys already at day old. Keeping flocks of different ages in parallel on a farm reduced the probability of becoming affected by histomonosis by half (OR = 0.461, P = 0.060).
In addition, not implementing restricted access to an affected flock in case of any illness nearly doubled the probability of an outbreak in the univariate, single-factorial analysis (OR = 1.990, P = 0.098). In the univariate, multi-factorial model this variable seemed to triple the probability of an outbreak (OR = 3.283, P = 0.009; Table 2).

Discussion
The study population consisted of a convenience sample as no random sampling from a preformed list of turkey farmers could be drawn. Furthermore, the calculated sample size could not be accomplished despite a suitable survey period and repeated advertisement of the study; hence, there were high P-values at all levels of analysis. However, the participation rate of 16.1% of all turkey farms, which harbour more than 1000 turkeys in Germany, can be considered sufficient for an epidemiological field study compared to previous studies (Campe et al., 2013;Abele et al., 2022). Moreover, it can be assumed that the dropout of the six participants, who stopped answering the questionnaire, did not create a selected study population. The same is valid for the dropout of the two intermediate-farms. We consider the sample size representative for the turkey farms in Germany as the geographical distribution of the participating farms was well balanced throughout Germany, including the  distribution concerning the density of turkey-keeping farms in a federal state (Supplemental Material 1). Based on the distribution of the study population between the control-group and the case-group, we cannot determine whether the region could be a risk factor or not.
Unanswered questions were documented in the response option "No data". Sensitivity analysis showed that no systematic refusal to answers could be detected within each group (Dohoo et al., 2003). Due to the small sample size, questionnaires with an unanswered question could not be excluded from the study population. As a result, the category "No data" was included in all evaluation procedures, but its results were not interpreted in a contentual manner.
The OR of remaining possible ways of pathogen introduction that are not listed in the result section could not be calculated (n.c.; n = 3; Table 1 and Supplemental Material 3) or did not indicate for any effect (n = 4). For 12 variables, the content could not be plausibly interpreted due to lack of variance. For following variables, reverse causation could explain the implausibility of the results, while considering that the questionnaire referred to the situation before a histomonosis outbreak had appeared. We may speculate that only farms with an outbreak are sensitized to the usage and change of disinfectants. This suggests that no usage or no change of disinfectant could protect against an outbreak. However, we assume that only case-farms changed the product regularly, because they had to experience an outbreak. The same reverse causation could be stated for the usage of equipment outside the turkey house.
Due to the large amount of analysed data, only notable results are discussed in the following paragraphs (factors marked in bold in Figure 1).
In addition to earthworms and the caecal worm Heterakis gallinarum as general hosts, H. meleagridis has also been detected in mechanical vectors, such as flies and beetles (McDougald, 2005;Hauck et al., 2010;Hauck & Hafez, 2013). The results of our study indicate that humus-rich soil may increase the probability for an outbreak of histomonosis by providing an attractive environment for earthworms (Lavelle, 1988;Cortez, 1998;Perreault & Whalen, 2006). The pathogen can persist in earthworms for years (Lund & Chute, 1973;Kemp & Franson, 1975;Piearce & Phillips, 1980). Other results suggested that unpaved driveways may provide an attractive environment for vectors. They may favour the introduction of H. meleagridis-infected vectors via footwear or machinery. The usage of emergency or maintenance doors or both to unfortified ground outside the turkey house without re-passing the hygiene lock may also allow a simple pathogen-introduction.
Despite a low frequency of insect and snail control, along with a generally low frequency of vector observations, our data suggest that the observation of earthworms, snails and/or beetles in the environment of the turkey house may increase the probability of becoming affected by histomonosis. To our knowledge, this was the first analysis of a potential association between the environmental observation of snails and the occurrence of histomonosis. As observation and actual occurrence of these vectors are not necessarily in line, a bias can be assumed caused by increased awareness of the study participants. Here, reverse causation may have influenced the results in the way that owners of histomonosis-affected farms may have been more aware of possible vectors for pathogen-introduction after a histomonosis outbreak than participants who did not previously have problems with the disease (Dohoo et al., 2003). Therefore, a noticeable increase in the density of the above-mentioned vectors might lead to increased biosecurity measures to prevent histomonosis on the respective farms.
H. meleagridis survives longer in a moist environment (Lotfi et al., 2012). Earthworms may withdraw into litter, increasing the potential for an infection (Lavelle, 1988;Cortez, 1998;Perreault & Whalen, 2006). We may speculate that closed turkey houses with an insufficient climate management may have a higher risk for increased litter humidity compared to open-type houses (Hermans et al., 2006). This hypothesis was substantiated by the finding of this study that closed houses doubled the probability for histomonosis on a farm. In addition, an irregular frequency of litter refill, for example by making the frequency dependent on weather conditions rather than a fixed schedule, may pose a possible risk for increased moisture in the house and litter material. Our data indicate that a weather-dependent litter management nearly tripled the probability for a farm to become affected by histomonosis. An irregular frequency of litter refill could also increase the contact with H. meleagridis-infected faeces and favour the route of cloacal infection (Kemp & Reid, 1966;Huber et al., 2006). The combination of an inappropriate frequency of litter refill and the usage of low absorbency litter material, such as straw, which was identified to increase the risk of an outbreak of histomonosis by a factor of nearly three as well, could increase the risk of higher humidity (Toledo et al., 2019;Diarra et al., 2021).
A pathogen introduction is possible through wild birds. In addition to pheasants and partridges, which can also suffer clinically from histomonosis, the pathogen and its vector Heterakis gallinarum have been detected in crows, as a species of song birds, which have been identified to increase the risk of an outbreak by approximately 75% in this study (Madsen, 1952;Kemp & Franson, 1975;McDougald, 2005;Eslami et al., 2007;Potts, 2009). Wild birds can introduce the pathogen into a flock by contaminating the straw with faeces, either freshly harvested on the field or in the barn while in storage. The introduction of fresh droppings of infected wild birds through wheels of machinery or footwear is also possible. Our study also suggests open storage of litter material and an inadequate cleaning and disinfection of the machine wheels before entering the house as risk factors for an outbreak of histomonosis.
Furthermore, a high density of poultry farms poses a possible risk of pathogen transmission, which can be justified by the susceptibility of poultry species other than turkeys to the pathogen H. meleagridis, which has been indicated in previous studies (Callait-Cardinal et al., 2006;Powell et al., 2009;Popp et al., 2011;Mitra et al., 2018;Lagler et al., 2019;Jones et al., 2020). In this study, the close proximity to broiler farms and backyard flocks nearly doubled the probability for becoming affected by histomonosis. Insufficient biosecurity measures for backyard flocks, as well as shared use of vehicles and equipment between farms in poultry-dense areas, may favour cross-contamination (Delpont et al., 2021;Huneau-Salaun et al., 2022).
The transport of 4-to 6-week-old turkeys to fattening farms could also lead to an increased probability for pathogen introduction, because transporting trucks could be contaminated with H. meleagridis, as shown for other pathogens (Huneau-Salaun et al., 2022). Furthermore, farms that kept several age groups in parallel seemed to be more sensitive to hygiene measures and, therefore, they have a lower probability of becoming affected by histomonosis than farms that have an all-in all-out regime. For this reason, a generally restricted access could reduce the risk of pathogen introduction into a flock by nearly two times.
Increased cleaning efforts after any case of illness seem to reduce the risk of a histomonosis outbreak, possibly due to the side effect that H. meleagridis may also be reduced in the environment not only through the reduction of animate and inanimate vectors but also its cyst-like stage (Munsch, et al., 2009;. It is not known how widespread this stage of H. meleagridis is in the environment and how long it survives without being ingested by vectors. An in vitro study proved that the pathogen survives up to 1 h on rubber material, e.g. on wheels of machinery, and 6 h in litter material (Lotfi et al., 2012). Further research is needed to understand more about the occurrence of the cyst-like stage of H. meleagridis in the environment, on its tenacity and the associated impact on the occurrence of histomonosis.
In conclusion, this retrospective case-control-study clearly suggests that, especially in poultry-dense regions, biosecurity measures are most important for the control of H. meleagridis infections in turkey production, which includes regular cleaning and disinfection of turkey houses as well as equipment. The introduction of potential vectors and wild birds, as a H. meleagridis-reservoir, should be avoided at all times. A screening of wild birds for H. meleagridis may be advisable to determine the prevalence of H. meleagridis and evaluate the associated potential risk for poultry. The climate conditions in the turkey houses should be monitored carefully and an adequate litter management should be carried out to avoid favourable conditions for the survival of the pathogen and its direct or mechanical vectors. Overall, this study provides important information on possible ways of pathogen introduction. Nevertheless, our findings should be substantiated by further research in other regions of the world with different poultry management practices and, if possible, with higher numbers of participants.