Assessing Campylobacter cross-contamination of Danish broiler flocks at slaughterhouses considering true flock prevalence estimates and ad-hoc sampling

Alessandro Foddai*, Nao Takeuchi-Storm, Birgitte Borck Høg, Jette Sejer Kjeldgaard, Jens Kirk Andersen, Johanne Ellis-Iversen

*Corresponding author for this work

Research output: Contribution to journalJournal articleResearchpeer-review

32 Downloads (Pure)


Campylobacter cross-contamination of Danish broiler flocks at slaughterhouses was investigated using data from two national surveillance components and from ad-hoc sampling. The animal level (AL) and food safety (FS) components from 2018 were compared. The AL component contained results of PCR on pools of cloacal swabs from 3,012 flocks processed at two Danish slaughterhouses (S1-S2), while the FS component regarded culture testing of leg skins from 999/3,012 flocks. The monthly “apparent” (AP) and “true” flock prevalence (TP) were estimated. Agreement between components was measured in percentage and in weighted-Kappa values. The relationship between the occurrence of cross-contamination (flock positive only in the FS component = cross-contaminated or CC, vs. flock negative in both components or NegBoth), slaughterhouse and surveillance period (quarter: Q1 to Q4) was evaluated by a generalized linear mixed effects (GLM) model. Thereafter, a linear mixed effects (LME) model was used to investigate the relationship between the level of meat contamination of carcass positive flocks (y = log10 colony forming units per gram, cfu/g), slaughterhouse, surveillance period, and flock type (CC vs. positive in both components or PosBoth). For both models, the farm was the random effect. Finally, in autumn 2019, ad-hoc field investigations were carried out testing caecal and neck skin samples, from two consecutive flocks at S1 and S2. Whole genome sequencing (WGS) was performed on isolates, for multilocus sequence typing (MLST) and single nucleotide polymorphisms (SNP) analysis. The monthly TP was always higher for the FS than for the AL component. Agreement between the components was substantial, but 8.1–8.6% of the flocks were CC. Those had median cfu/g 21–28 times lower than that of PosBoth flocks. In the GLM model, the explanatory variables were both significant (P-value <0.05). For example, the odds ratios (ORs) were 8.4 (95% CI: 4.0; 17.6) for Q3 vs. Q1, and 3.1 (1.8; 5.2) for S2 vs. S1. In the LME model, the flock type and the interaction between the other two variables, were significant. In the field study, a caecal positive flock was succeeded by an initially negative flock, in one out of five sampling sessions at S2. The cecal negative flock was positive in 58.3% of the neck skins with the isolate genetically similar to that from the caecal positive flock. Those results show that cross-contamination can be affected by surveillance periods and slaughterhouses, and it can contribute significantly to the TP of carcass positive flocks.
Original languageEnglish
Article number100214
JournalMicrobial Risk Analysis
Number of pages10
Publication statusPublished - 2022


  • Campylobacter spp.
  • Data integration
  • Prevalence
  • Cross-contamination
  • Field investigation


Dive into the research topics of 'Assessing Campylobacter cross-contamination of Danish broiler flocks at slaughterhouses considering true flock prevalence estimates and ad-hoc sampling'. Together they form a unique fingerprint.

Cite this