No association between partial depopulation and Campylobacter spp. colonization of Dutch broiler flocks

Authors


  • Present address: A.D. Russa, Department of Anatomy and Histology, Muhimbili University College of Health Sciences, Dar es Salaam, Tanzania.

A. Bouma, Department of Farm Animal Health, Faculty of Veterinary Medicine, Utrecht University, Yalelaan 7, 3584 CL Utrecht, the Netherlands (e-mail: a.bouma@vet.uu.nl).

Abstract

Aims:  To determine whether an association exists between partial depopulation of a flock and increased Campylobacter colonization in that flock.

Methods and Results:  Data from 1737 flocks of two Dutch integrators were used. Flocks that experienced partial depopulation were defined as ‘exposed’ and those that did not as ‘nonexposed’. Multivariable modelling was accomplished with, in addition to ‘exposure’, the independent variables ‘age of broilers at slaughter’ and ‘season’ to adjust for possible confounding. The response variable was ‘Campylobacter colonization’. The odds ratio (OR) for partial depopulation for integrator A was 0·8 [95% CI (0·4, 1·8)]; for integrator B the OR = 0·8 [95% CI (0·5, 1·3)]. Age and season were confounders: the difference in Campylobacter status between exposed and nonexposed flocks of integrator A could be explained by both variables; for integrator B, only season was associated with Campylobacter status.

Conclusions:  We found no significant association between partial depopulation and an increased risk of Campylobacter colonization among broiler flocks at final depopulation.

Significance and Impact of the Study:  This study demonstrates that Campylobacter colonization in a broiler flock is not influenced by the partial depopulation of that flock.

Introduction

Campylobacter species are a major cause of food-borne gastrointestinal infections in humans (World Health Organization 2000). An important source of human Campylobacter infections is the handling and consumption of undercooked contaminated chicken meat (e.g. Kapperud et al. 1992; Eberhart-Phillips et al. 1997; Evans et al. 1998; Allos 2001; Neimann et al. 2003). Consequently, an important contribution to the reduction of the exposure of humans could be obtained by a reduction of Campylobacter colonization of broiler flocks.

Although the exact contamination sources and transmission routes of Campylobacter are still unknown, many risk factors with respect to increased Campylobacter colonization have been identified (e.g. Kapperud et al. 1992, 1993; Jacobs-Reitsma et al. 1994; Van de Giessen et al. 1996; Jacobs-Reitsma 1997; Refregier-Petton et al. 2001; Bouwknegt et al. 2004). These include, for example, poor farm hygiene, short time between two consecutive-production periods, the presence of other farm animals, rodents and insects, seasonal variations and partial depopulation. Although biosecurity measures, such as hygienic barriers and rodent control, have resulted in a decreased colonization of Campylobacter on broiler farms (Gibbens et al. 2001), infections on broiler farms still occur frequently that necessitates further investigation of the role of risk factors.

Partial depopulation (or batch depopulation) has been reported as risk factor for Campylobacter colonization of broilers (Hald et al. 2000, 2001; Jacobs-Reitsma et al. 2001). Partial depopulation is a process where a small part of early market-weight attaining broilers is slaughtered, followed by complete depopulation of the remaining flock, c. 1 week later. Because of the economic benefit, it is practised by several farmers and slaughter companies.

As mentioned before, it has been suggested that the catching teams might introduce Campylobacter spp., when going from farm to farm (Kapperud et al. 1993; Jacobs-Reitsma et al. 2001). However, these conclusions are not necessarily justified by these studies. The findings of Jacobs-Reitsma et al. (2001) were based on a postal questionnaire, a study type that is often hampered by nonresponse and recall bias (Thrusfield 1997). Moreover, the study was designed as case–control study, in which the risk factor and Campylobacter-colonization were established at the same time. This implies that the principle of time sequence of events to establish a possibly causal association (Thrusfield 1997) was not met. Also, Hald et al. (2001) noted that partial depopulation increased the number of Campylobacter spp. colonized broiler flocks. However, this study only reported that flocks that were Campylobacter negative at partial depopulation were positive 1 week later and did not include control flock that were not partially depleted. Other studies (Hald et al. 2000; Wedderkopp et al. 2000) did not take into account confounders that have also been reported to be associated with increased infections, such as birds’ age at slaughter (Berndtson et al. 1996; Evans and Sayers 2000; Bouwknegt et al. 2004) and seasonal variation (Willis et al. 1991; Kapperud et al. 1993; Jacobs-Reitsma et al. 1994; Berndtson et al. 1996; Willis and Murray 1997; Refregier-Petton et al. 2001; Newell and Fearnly 2003; Bouwknegt et al. 2004).

Therefore, these studies do not allow to decide as to whether partial depopulation is a cause of introduction of Campylobacter on broiler farms or not. As the impact for farmers of ending the practice of partial depopulation might be large, it should be investigated properly whether or not partial depopulation is really a risk factor, or that the increased risk can be explained by confounders like age at final depopulation or season.

Therefore, the aim of this study was to investigate whether and to what extent partial depopulation of broiler flocks increased the risk of Campylobacter colonization of the finally depleted flock. The study design was a cohort study, which has the advantage that the risk factor is registered before the Campylobacter colonization was established. Moreover, we took into account other factors likely to be associated with Campylobacter colonization, such as age of the broilers and season. We used historical data from two large Dutch poultry integrators gathered for the Dutch National Campylobacter monitoring programme.

Materials and methods

The Campylobacter monitoring programme

Since 1997, a monitoring programme for Campylobacter colonization of Dutch broiler farms has been carried out by the Dutch Product Board of Livestock and Meat (PVE). Twice a year, 30 caecal samples from broiler flocks were collected from the slaughterhouse and pooled. Caecal samples were collected from the slaughterhouses as it could be carried out in a better-controlled way than faecal samples taken at the farm, and, in addition, the concentration of bacteria in caecal samples is generally higher than in faeces (Achen et al. 2004; Rudi et al. 2004). The sample size allowed detection of a flock when the minimal prevalence is 10% (Thrusfield 1997), a prevalence that is usually obtained within a few days after introduction of Campylobacter (e.g. Annan-Prah and Janc 1988; Jacobs-Reitsma et al. 1995). The samples were pooled into one sampling bag or tube and sent by regular mail (without special treatment like shipment on ice) to the laboratory, generally within 1–2 days after sampling. The samples were examined for the presence of Campylobacter by qualified laboratories of the PVE. In short, this included direct streaking of pooled material onto a Campylobacter-selective charcoal cefoperazone desoxycholate agar plate, followed by micro-aerobic incubation at 42°C for 48 h. The test results were reported to the slaughterhouse management to be registered in a database, because the slaughterhouse is responsible for reporting data to the PVE.

Study population

In this study, we used data from two large Dutch poultry integrators (Integrators A and B). The available data were based on lots (i.e. a group of broilers originating from one farm and transported to a slaughterhouse on 1 day). As one lot could contain birds from more than one flock (i.e. a group of broilers raised in one shed in one production cycle), we used only those lots that originated from a single flock, because otherwise the Campylobacter-colonization could not be linked to a particular flock. Flocks that had experienced partial depopulation once were defined as ‘exposed’ and those that did not as ‘nonexposed’.

Information about the flocks included in this study is shown in Table 1. The following information was available at the flock level: dates of stocking, partial depopulation and final depopulation, number of broilers, and the Campylobacter colonization at final depopulation. The dates of partial depopulation in integrator B, however, had not been recorded, only whether it had been practised or not.

Table 1.  Overview of the technical data of integrators A and B (summary statistics)
 Integrator
AB
No. of flocks839898
No. of farms329178
Study periodJanuary to December 2000January 2000 to December 2001
No. of farms with
 Nonexposed flocks only2749
 Exposed flocks only18122
 Both exposed and nonexposed flocks3747
Age at slaughter (days)
 Nonexposed flocks34–5244–53
 Exposed flocks38–5040–52

Analysis of data

For the statistical analysis, we only used the results of the caecal tests of the flocks at final depopulation. We would have preferred to analyse the rate at which flocks became positive between partial depopulation and final depopulation, because this would allow a better technique for assessing risk and identifying causes. However, the available data made this impossible. The reason was that the status of partially depopulated flocks was often unknown, because either no faeces samples had been collected, or the corresponding lot comprised more than one flock. Moreover, for obvious reasons, no faeces samples had been collected in the nonexposed group at the time of partial depopulation of exposed flocks, making a comparison impossible.

We used the limited available data of integrator A at the time of partial depopulation to get an indication for a possible difference in Campylobacter colonization between exposed and nonexposed flocks at the age of partial depopulation of the first. For that purpose we established the prevalence of Campylobacter-colonized exposed flocks at partial depopulation at 34–40 days and the prevalence of Campylobacter-colonized nonexposed flocks that had been finally depopulated at that same age.

We statistically analysed the data of integrators A and B separately, because the different age distributions at final depopulation made it impossible to properly correct for age in a combined analysis. Moreover, as age was assumed to be a confounder, it was necessary for interpretation of the result that the analysis of the data of each integrator was based on comparable age groups in exposed and nonexposed flocks. Therefore, only data from flocks slaughtered between 41 and 47 days for integrator A and between 45 and 50 days for integrator B were included. In addition, we distinguished winter (January to March), spring (April to June), summer (July to September) and autumn (October to December), and categorized the flocks with respect to season on the basis of the date of final depopulation.

A logistic regression model was fitted, in which farm was included as random effect, because more flocks from one farm could be included in the data set. Multivariable modelling was accomplished using R version 1.7.1 (Ihaka and Gentleman 1996; http://www.r-project.org). The independent variables were: exposure, age and season in which birds were depleted. The response variable was the Campylobacter colonization of the flocks, based on caecal sample, at the time of final depopulation.

The following starting model was used:

image

where π is the probability of being infected at final slaughter. β1,β2 and β3 are the ln odds ratio (OR) for respective variables; ɛ is the error term. Throughout the modelling process, the final model was obtained by backward elimination procedure based on Akaike Information Criteria (AIC) (Burnham and Anderson 1998).

Results

Descriptive statistics

Technical data about the two integrators are shown in Table 1.

Integrator A.  The prevalence of Campylobacter-positive flocks in integrator A was 55·7% in the exposed and 34·0% in the nonexposed group (Table 2). However, the average age at sample collection was 4 days older in the exposed than in the nonexposed flocks. At partial depopulation 41% of the exposed flocks were Campylobacter-positive, whereas 26% of the nonexposed flocks were positive at final depopulation at that same age (34–40 days). Table 2 shows that the proportion of Campylobacter-positive flocks increased with age. Moreover, the proportion of Campylobacter-positive flocks was higher in autumn than in the other seasons (Table 3).

Table 2. Campylobacter colonization of the flocks of integrators A and B based on caecal samples
IntegratorAge range† (days)ExposureNo. Campylobacter colonized flocks*Mean age at slaughter (days)No. of flocks
Positive (%)Negative (%)
  1. *Number of flocks (percentage of flocks).

  2. †Data from all flocks and for those flocks used in the statistical analysis.

  3. ‡Data from all flocks.

  4. §Data from flocks of comparable age, used for the statistical analysis.

A34–52‡Exposed59 (55·7)47 (44·3)43·6106
Nonexposed249 (34·0)484 (66·0)39·6733
34–40§Exposed44 (41·5)62 (58·5)36·9106
Nonexposed114 (26·1)323 (3·9)37·1437
41–47§Exposed55 (56·1)43 (43·9)43·698
Nonexposed127 (45·2)154 (54·8)43·2281
B44–53‡Exposed345 (44·9)423 (55·1)47·5768
Nonexposed58 (44·6)72 (55·4)47·4130
45–50§Exposed319 (43·8)409 (56·2)47·7728
Nonexposed57 (45·6)68 (54·4)47·4125
Table 3.  Proportion of Campylobacter colonized flocks per season for integrators A and B
SeasonPercentage positive flocks
Integrator AIntegrator B
Exposed flocksUnexposed flocksExposed flocksUnexposed flocks
Winter50·034·850·437·5
Spring52·037·264·966·7
Summer52·850·748·039·1
Autumn68·055·345·843·8

Integrator B.  The percentage of Campylobacter-positive flocks was 44·9% of the exposed and 44·6% of the nonexposed flocks (Table 2). The average age at final depopulation was almost the same in the exposed and the nonexposed flocks. Moreover, no increase of Campylobacter-positive flocks with age was observed. Most flocks were Campylobacter-positive in summer and autumn (Table 3).

Multivariable analysis

Age at slaughter of the flocks varied substantially. Therefore, comparable age group were made to allow an adequate statistical analysis. Data of these subsets are shown in Table 2.

Integrator A.  The multivariable analysis of flocks, slaughtered at 41–47 days of age, showed no significant association between partial depopulation and Campylobacter colonization [OR = 0·8, 95% confidence interval (CI) (0·4, 1·8)]. Moreover, the AIC indicated that the model with only age and season fitted the data better than the model that also included exposure. The OR for age was 1·7 (1·4, 2·0). Summer and autumn were associated with higher number of positive flocks compared with the first half year: OR = 2·0 [95% CI (1·1, 3·8)] and OR = 3·7 [95% CI (1·9, 7·3)] respectively.

Integrator B.  Also in integrator B multivariable analysis of flocks, slaughtered at 45–50 days of age, did not show a significant association between partial depopulation and Campylobacter colonization [OR = 0·8, 95% CI (0·5, 1·3)]. Again, the model without exposure fitted the data better than the model that included exposure. The OR for age was 1·0 (0·9, 1·1). Moreover, Campylobacter colonization was significantly lower in spring than in the other seasons [OR = 0·6, 95% CI (0·4, 0·8)].

Discussion

In this study we did not find an association between partial depopulation and Campylobacter colonization of broiler flocks at final depopulation. However, in integrator A age and depopulation in summer or autumn were associated with an increased risk of Campylobacter colonization. In integrator B, depopulation in spring was associated with a decreased risk of Campylobacter colonization.

Our results are in conflict with earlier reports (Hald et al. 2000, 2001; Wedderkopp et al. 2000; Jacobs-Reitsma et al. 2001) that indicated partial depopulation as a risk factor for Campylobacter colonization. Although this difference could be partly due to different hygienic procedures of the catching crews, it is likely that confounding also played a role. The reason is that the studies mentioned above did not take age into account as a possible confounding variable. The results of integrator A demonstrate that the association between partial depopulation and Campylobacter colonization might disappear when age and season are taken into account. The explanation for age being a confounder is that the longer the birds are kept on farm the higher the chance of colonization of the broilers (Berndtson et al. 1996; Evans and Sayers 2000; Bouwknegt et al. 2004). As exposed flocks of integrator A were kept at the farm longer than the nonexposed flocks, partial depopulation was indirectly associated with increased risk of Campylobacter colonization. This explanation is further supported by the results obtained in integrator B, where final depopulation of the exposed and nonexposed flocks took place at approximately the same time and no difference in the prevalence of Campylobacter colonization was observed. Season was a confounder for both integrators, which is consistent with other findings on broiler flocks and in humans (Patrick et al. 2004; Louis et al. 2005). Until now, no explanation for this finding has been described.

Our findings show no evidence that partial depopulation itself increases the colonization of Campylobacter in broiler flocks, as the higher incidence in exposed flocks can be explained by the older age of the broilers at slaughter. However, this finding is subject to the assumption that the exposed and nonexposed flocks were comparable with respect to farm management and hygiene and Campylobacter colonization at the time of partial depopulation. We tested the first assumption by investigating the association between partial depopulation and Campylobacter colonization on farms that included exposed and nonexposed flocks. Analysis of this subset of the database also failed to find an association (data not shown), indicating no major influence of differences in management and hygiene. Moreover, the data in Table 2 suggest that the lack of association does not result from a low prevalence of Campylobacter colonization at the time of partial depopulation. However, definitive proof of the validity of these assumptions would require a field study in which partial depopulation is randomly allocated to flocks.

We have shown that routinely collected data can be of use for an analysis of the association between partial depopulation and Campylobacter-colonization of broiler flocks. Although the results of this study have limitations, our findings suggest that ending the practice of partial depopulation would not reduce the number of Campylobacter-colonized flocks at final depopulation.

Ancillary