Risk factors associated with colibacillosis outbreaks in caged layer flocks

Abstract

Colibacillosis appears to be of increasing significance in layer flocks, but there have been no studies of the risk factors associated with outbreaks. This study aimed to investigate the possible associations between risk factors of non-infectious nature and outbreaks of mortality due to colibacillosis in flocks of caged layer hens. Information on management, biosecurity measures and housing conditions was collected in 20 flocks suffering from the disease and in 20 clinically healthy control flocks. The data were processed using multiple logistic regression. The statistical analysis demonstrated that an increase in the distance to the nearest poultry farm by 1 km was associated with a six-fold decreased risk of an outbreak of colibacillosis (odds ratio=0.16). Furthermore, a 1 l increase in cage volume per hen was associated with a 33% decrease in the risk of an outbreak (odds ratio=0.75). It was concluded that the distance between poultry farms and the hen density in the cages are important risk factors for outbreaks of colibacillosis in flocks of layer hens.

Introduction

Colibacillosis, caused by avian pathogenic Escherichia coli (APEC), is one of the main causes of economic loss in the poultry industry worldwide (Morris, 1989; Dho-Moulin, 1993; Yogaratnam, 1995; Elfadil et al., 1996; Barnes & Gross, 1997). Traditionally, it was mostly associated with losses in broilers, but in Europe from the mid-1990s onwards it has been increasingly observed in layer hens (Zanella et al., 2000; Vandekerchove et al., 2004a). In our recent field study, we described a distinct syndrome associated with colibacillosis in laying hens, characterized by acute mortality without prior clinical signs of disease and without a significant impact on egg production or quality. At necropsy, the layers had lesions compatible with colisepticaemia (polyserositis, including perihepatitis, pericarditis and peritonitis with yolk material deposited in the peritoneal cavity). The majority of the outbreaks occurred around the period of peak egg production, and cumulative mortality sometimes exceeded 9%. In the majority of the flocks, most of the E. coli isolated belonged to the O78 serotype, although other serotypes, including O2 and O18, were also detected. The strains usually possessed F11 fimbriae and flagella (Vandekerchove et al., 2004a). Infection with other pathogens is reported to make chickens more susceptible to colibacillosis (Barnes & Gross, 1997). However, our earlier studies found no significant associations between colibacillosis outbreaks and serum antibodies against infectious bronchitis virus (IBV), Newcastle disease virus (NDV), avian pneumovirus (APV), Mycoplasma gallisepticum, Mycoplasma synoviae or Ornithobacterium rhinotracheale (Vandekerchove et al., 2004b). These results indicate that, even though in some cases colibacillosis can be triggered by other infectious agents, APEC may act as primary pathogens.

There is no information about the possible associations between risk factors of a non-infectious nature and outbreaks of mortality due to APEC in flocks of caged layer hens. Several potential risk factors are described in the literature. Rodents may be carriers of APEC (Barnes & Gross, 1997). Water hygiene is also important. Dhillon & Jack (1996) reported that adding chlorine to the drinking water was effective in controlling the spread of disease and in reducing mortality in two farms with outbreaks of colibacillosis. Feed composition may be of importance, as van Harn & van Middelkoop (2000) reported a reduction in mortality associated with E. coli in broilers when formic acid was added to the feed. Colibacillosis is also believed to be secondary to unfavourable housing conditions (Barnes & Gross, 1997). Stress increases susceptibility to disease in general (Freeman, 1976), and to colibacillosis in particular (Leitner & Heller, 1992). In general, flies and other insects may play a part in disease introduction and spread (Zander et al., 1997). Dermanyssus gallinae may cause stress, their feeding can eventually lead to anaemia, and they may act as vectors (Axtell & Arends, 1990). Frequent removal of dead hens from the house is important to control the spread of disease (Zander et al., 1997). High egg production is associated with an imbalance between the oestrogen and the progesterone levels, which causes reduced resistance to ascending infections of the oviduct (Kohlert, 1968). When hens are beak trimmed at a young age, their beaks may regrow, and pecking can lead to cannibalism (Riddell, 1997), allowing infection through the compromised skin or mucosal barriers (Barnes & Gross, 1997). However, no reports are available about the importance of these factors in layer hens, with the exception of the early report by Kohlert (1968).

The aim of this study was to investigate the possible associations between hygiene and biosecurity measures, flock health, management factors, infrastructure, farm situation, production parameters and rearing management, and increased mortality associated with E. coli in layer flocks. For this purpose, a matched case–control study was performed in 40 caged layer flocks.

Materials and Methods

Flocks

The study was run from February 2001 to July 2002. It was set up as a case–control study in 40 commercial caged layer flocks on 25 different farms in the northern part of Belgium. Twenty flocks suffering from mortality associated with colibacillosis were matched with 20 clinically healthy control flocks of the same ages. Seven of the control flocks and 17 of the affected flocks were situated in the north-western part of Belgium (provinces of West and East Flanders). Thirteen of the control flocks and three of the affected flocks were situated in the north-eastern part of Belgium (provinces of Antwerp and Limburg). On 14 farms, one flock was sampled. On eight farms two flocks were sampled, on two farms three flocks were sampled and on one farm four flocks were sampled. Four farms contained both a colibacillosis and a control flock. All farms were multi-age farms. Ages of the control flocks varied between 21 and 74 weeks, while the ages of the affected flocks ranged from 21 to 87 weeks. Flock sizes ranged from 11 005 to 50 000. Breeds were Isa Brown (27/40), Lohman Selected Leghorn (6/40), Lohman Brown (3/40), Hisex Brown (3/40) or Hisex White (1/40).

The 20 affected flocks were identified on the basis of the criteria of Vandekerchove et al. (2004a): (1) a ‘farmer-reported’ increase in mortality, as compared with the normal baseline mortality in the flock; (2) detection of lesions compatible with colibacillosis after necropsy of five hens; and (3) E. coli isolated from heart and/or liver in pure or abundant cultures according to the criteria of Vandekerchove et al. (2004a). A control flock was defined as a flock in the house of which no increased mortality, compared with the standards of the breeding organization, had been observed during the current and the previous batch(es). The study was based on voluntary participation. Flocks suffering from an outbreak of colibacillosis were visited within 2 weeks of the onset of the increased mortality to collect diseased or dead hens for necropsy (Vandekerchove et al., 2004a). This was followed by collection of four sets of blood samples at 2-week intervals to examine the flocks for possible seroconversions against IBV, NDV, APV, M. gallisepticum, M. synoviae or O. rhinotracheale (Vandekerchove et al., 2004b). Data collection for the risk factor analysis was performed at the fourth blood sampling.

As soon as an affected flock was identified, a control flock of comparable age was sought. Control flocks were sampled and analysed in the same way as the affected flocks.

The maximum weekly mortality ranged from 0.07 to 0.30% in the control flocks and from 0.26 to 1.71% in the affected flocks. Increased mortality in the affected flocks lasted from 3 to 10 weeks, with a three-fold to eight-fold increase in the number of dead hens per week within a 1-week to 3-week period following the onset of the outbreak. A decrease in egg production exceeding 2% was seen in one control flock and in six affected flocks. A decrease in egg quality (percentage of second-grade eggs exceeding 1.5%) was seen in three affected flocks. In only one affected flock was a decrease in both egg production and egg quality seen (Vandekerchove et al., 2004a). No significant associations were seen between the colibacillosis outbreaks and antibodies against IBV, NDV, APV, M. gallisepticum, M. synoviae or O. rhinotracheale (Vandekerchove et al., 2004b).

Data collection

The farm managers were asked to answer a questionnaire during a face-to-face interview. The questionnaire was designed to collect data on biosecurity measures, flock health, management factors, infrastructure, farm situation and production parameters. This questionnaire was drawn up to assess risk factors for colibacillosis outbreaks suggested in the literature, by poultry veterinarians and poultry technicians, and was tested on farm managers and poultry veterinarians before use in this study. All interviews were conducted by the first author of this paper. Data on vaccinations and age of debeaking were obtained from the breeding and rearing companies.

The score for biosecurity measures was calculated as follows. Changing shoes and clothes to enter the hen house, washing hands before entering the hen house, and other possible measures were each given two points when always applied, one point when usually applied, and zero points when never applied. These data were tested individually. They were also tested as a total score for the farm manager, and as a total score for the visitors. Moreover, the overall biosecurity score was calculated by adding the farmer's total score to the total score for visitors divided by four, as the frequency of visitors was very low for all flocks.

Data validation

The questionnaires were encoded onto data sheets, where they were examined for errors and missing or unrealistic values. Errors were corrected after telephone contact with the farm manager or the breeding/rearing company. Unrealistic values were considered missing. Variables for which the observations could not be corrected or completed were excluded from the analyses.

In total, data on 144 variables were obtained. For some variables there was little or no variation, and for others too many classes with too few observations per class were obtained, so these variables were excluded from the analyses. Other variables were region dependent, as farmers in the north-western and north-eastern region were clients of different breeding/rearing companies, feed suppliers and housing equipment companies, resulting in the exclusion of most variables related to vaccination schemes, feed suppliers and infrastructure such as feed dispensing and watering systems. A total of 46 independent variables could be used for model building.

Data analysis

Conditional logistic regression (Genmod procedure; SAS, version 8.02) was used, with the farm identity as a random factor, to examine possible associations between the type of flock (i.e. control or affected flock), and the 46 variables described in Tables 1–4. The region of origin (north-western Belgium versus north-eastern Belgium) was forced into the model as a fixed effect to cope with the possible confounding effect of this variable. A forward stepwise procedure was used in view of the large number of variables available. At each step in the regression, all variables were tested, and the one giving the smallest P value in the likelihood ratio test, without causing multicollinearity with a previously selected variable, was retained. The final model was tested for interactions between the independent variables.

Table 1. Potential risk factors for colibacillosis outbreaks in caged layer flocks: descriptive data of categorical variables, significant in the bivariate analysis

Variable	Control flocks	Affected flocks	Bivariate P value
Type of animal with access to hen house			0.02
None	3	2
Cats	6	1
Rodents	7	13
More than one species	4	4
Regular treatment against flies			0.02
No	9	7
Yes	11	13
Pattern light increase at the beginning of the batch			0.0003
More than 30 min per week	7	12
30 min per week	13	8
Pre-lay feed offered before layer feed at the beginning of the batch			0.005
No	8	17
Yes	12	3
Number of other poultry farms (commercially kept layers or broilers) within a 1 km radius			0.005
None	20	15
1 to 3	0	4
More than 3	0	1
Percentage in lay at 22 weeks versus target			0.005
Below the target	10	13
On the target	6	1
Above the target	4	6

Table 2. Potential risk factors for colibacillosis outbreaks in caged layer flocks: descriptive data of continuous variables, significant in the bivariate analysis

	Mean (minimum to maximum)
Variable	Control flocks	Affected flocks	Bivariate P value
Biosecurity measures
Visitors entering hen house (n/month)	0.8 (0 to 2)	1.4 (0 to 4)	0.004
Frequency of water disinfectant use (n/year)	43.4 (0 to 365)	84.3 (0 to 365)	0.04
Infrastructure
Number of hens in the flock (n)	26 996 (11 005 to 44 000)	36 157 (23 400 to 50 000)	0.0003
Well depth (m)	82.2 (7 to 250)	181.2 (18 to 300)	0.04
Farm situation
Distance to the nearest poultry farm (km)	3.3 (0 to 10)	1.0 (0 to 3)	0.0006
Rearing period
Age of beak trimming (days)	25.4 (7 to 42)	13.6 (9 to 42)	0.03

Table 3. Risk factors for colibacillosis outbreaks in caged layer flocks not found significant in the bivariate analysis: categorical variables

Variable
Hygienic preventive measures
Dead hen removal from the house (daily or not daily)
Use of contract cleaners between batches (yes or no)
House cleaning method between batches (high pressure hose, compressed air, broom, or combination)
Disinfectant used on house between batches (formalin, commercial combination products containing quaternary ammonium compounds, aldehydes and alcohol, or other)
Cleaning of silos (using water, compressed air or broom, or silos never cleaned)
Frequency of cleaning silos (never, less than every batch, or at least once per batch)
Disinfectants used as water additive (yes or no)
Flock health
Previous illness in other contemporary flock (other disease than colibacillosis, colibacillosis, or none)
Management factors
Use of feed supplements (extra vitamins, minerals) (never, sometimes or always)
Duration house empty between two batches (1 to 2 weeks or more)
Only poultry kept on the farm (yes or no)
Infrastructure
Drinking nipples shared between cages (yes or no)
Water mixing system (dosing pump or other)
Ventilation system (natural ventilation, fans, or a combination of both)
Fan orientation (natural ventilation only, width, or length)
Manure drying system (within the house or none within the house)
Farm situation
Distance from farmer's house to hen house (≤50 m or >50 m)
Production parameters
Percentage in lay at sampling (below, on, or above the target)
Rearing period
Extra vaccinations (E. coli, Salmonella, M. gallisepticum and/or avian pneumovirus) besides ‘standard scheme’ (yes or no)
Salmonella vaccine administered (yes or no)

Table 4. Risk factors for colibacillosis outbreaks in caged layer flocks found not significant in the bivariate analysis: continuous variables

Variable	Units
Biosecurity measures
Treatment frequency against D. gallinae	n/year
Frequency of dead hen removal from the house	n/week
Score on biosecurity measures	–
Management factors
Age at delivery	weeks
Light regime	h/day
Feeding frequency	n/day
Manure removal from the house	n/week
Infrastructure
Number of flocks on the farm	n
Total number of hens on the farm	n
Hen density in the house	n/m^3
Cage space per hen	l
Battery floors	n
Age of the hen house infrastructure	years
Rearing period
Vaccinations against Salmonella	n

Correlation between the independent variables was calculated using Pearson's correlations or the Spearman rank test.

Results

Data description

Tables 1–4 show the descriptions of the 26 categorical and 20 continuous variables that were used to build the model. In the bivariate analysis, the type of animal having access to the hen house (P=0.02), regular treatment against flies (P=0.02), the pattern of light increase at the beginning of the batch (P=0.0003), pre-lay feed offered before laying feed at the beginning of the batch (P=0.005), the number of other poultry farms within a 1 km radius (P=0.005), and the percentage in lay at 22 weeks versus the target (P=0.005) were the categorical variables significantly associated with the outcome variable (Table 1). The number of visitors entering the hen house (P=0.004), the frequency of water disinfectant use per year (P=0.04), the number of hens in the flock (P=0.0003), well depth (P=0.04), the distance to the nearest poultry farm (commercial layers or broilers) (P=0.0006), and the age of beak trimming (P=0.03) were the significant continuous variables in the bivariate analysis (Table 2). None of the biosecurity scores (individual or in combination) were significant. The risk factors that were not significant in the bivariate analysis are presented in Tables 3 and 4.

Data analysis

The only variables that were significantly associated with colibacillosis outbreaks in the multiple regression model, explaining a maximum of variation, were distance (km) to the nearest poultry farm, and the space per hen (l) in the cages, both of them continuous variables. The model is described in Table 5. The risk for a colibacillosis outbreak decreases sixfold with 1 km increase in distance to the nearest poultry farm (odds ratio=0.16). An increase in space of 1 l per hen in the cages decreases the risk for an outbreak by 33% (odds ratio=0.75).

Table 5. Significant risk factors in the final model^a

Variable	Odds ratio (95% confidence interval)	P value^b
Distance to nearest poultry farm (km)	0.16^c (0.05 to 0.55)	0.0035
Volume in cage (l/hen)	0.75^d (0.63 to 0.89)	0.0010
^a This model yielded a deviance of 16.11 from the null model (degrees of freedom=2, P=0.00032).
^b P values for the estimates of the odds ratios.
^c Increase in the distance by 1 km.
^d Increase in the volume per hen by 1 l.

Discussion

The objectives of this study were to identify and quantify the association between certain risk factors and colibacillosis outbreaks. All 46 variables were entered in a forward stepwise logistic regression, allowing the selection of the best multiple regression model. This model included the risk factors distance to the nearest poultry farm and the space per hen in the cages, both continuous variables.

Only a limited number of flocks participated in this study. The sampling was performed on an entirely voluntary basis, and farm managers, especially those whose flocks are performing well, restrict the number of visitors to a minimum to keep the risk of disease introduction low. For this reason, the case:control ratio was limited to 1:1. This reduced the power of the study, and perhaps some important variables have been discarded because the study was too small. Nevertheless, both categories contain flocks representative of the Belgian layer sector, and the sampling of 40 flocks allowed some significant conclusions. This study may be useful as a basis for more extensive studies in the future.

Little information is available on the importance of distances between farms for bacterial diseases. The farm density in an area implies the possibility of increased direct and indirect contacts, facilitating disease introduction. Even though the variable ‘distance to the nearest poultry farm’ cannot be simply translated to poultry farm density in an area, both a smaller distance and a higher farm density offer possibilities for increased contacts between the farms. For APEC transmission to caged hens, indirect contact through vectors is likely to be important. People can function as a vector, as can fomites, rodents, chicken mites or flies.

The second risk factor for colibacillosis outbreaks included in the model was the cage space per hen. A higher hen density per cage induces stress and hampers the maintenance of an optimal housing climate. Stress and unfavourable housing conditions are known to be predisposing factors for colibacillosis outbreaks (Leitner & Heller, 1992; Barnes & Gross, 1997). An increased housing density was found to increase the caecal numbers of Salmonella enteritidis in chicks (Asakura et al., 2001), which results in a higher infection pressure in the house. A similar effect might exist for APEC. A cost–benefit analysis could be performed to determine the optimal density in the cages, ensuring maximal economic return and minimal risk of disease outbreaks. However, there is also an increasing public concern for welfare issues in animal husbandry. As a result, European laws are coming into effect allowing more cage space per hen, which may prove to have a beneficiary effect on the layers’ health.

Several possible risk factors for colibacillosis or other disease outbreaks, including pest control, hygiene and biosecurity measures, have been described (Kohlert, 1968; Freeman, 1976; Axtell & Arends, 1990; Dhillon & Jack, 1996; Barnes & Gross, 1997; Riddell, 1997; Zander et al., 1997; van Harn & van Middelkoop, 2000). In the final multiple regression model only two variables were retained. Nevertheless, these other factors may play a part in flock health and should therefore be carefully controlled.

Earlier results indicated that colibacillosis in layers does not necessarily need priming by other respiratory pathogens. The model we obtained in this study demonstrated the importance of two non-infectious risk factors in initiating outbreaks of colibacillosis: the distance to the nearest poultry farm, and the hen density in the cages. This indicates that reducing the concentration of poultry farms in an area and the density of hens on the farms may be the most significant ways to control colibacillosis.

Acknowledgements

The authors wish to thank the farm managers, breeding and rearing companies, feed firms, poultry technicians and veterinary practitioners for their valuable help in making this study possible. Special thanks to J. Zoons, M. Dwars, A. Pijpers and Y. Schukken for helpful discussions.