Table 1.
Number of isolates in published human (h) and source (s) datasets and (last column) bootstrapped similarities of the human data with the human data in the NL1 dataset.
Figure 1.
Most common STs in human and chicken isolates in The Netherlands in two time periods.
Only the contributions of those CCs including STs that were found in the human data of 2002–2003 in proportions over 0.01 are represented. The contributions of less frequent STs within these CCs are summed and presented by the “+” symbol; the contributions of other CCs are omitted. For the human data of 2002–2003 the presented CCs make up for 83% of all isolates. For the chicken data of 2000–2007 and the human and chicken data of 2010–2011, these CCs make up for 65%, 67% and 52% of all data, respectively.
Figure 2.
Similarity of STs in chicken and in human isolates from samples collected in different years.
The x-axis gives the absolute difference between years in which the isolates from human cases and chicken were obtained. To enhance the size of the sample subsets, chicken isolates collected between 2000 and 2004 were aggregated and assigned to be collected in 2002, those collected between 2005 and 2007 were assigned to be collected in 2006, and those collected in 2010–2011 were assigned to be collected in 2010. Human isolates were arranged in three groups: 2002, 2003, and 2010–2011. The y-axis represents the PSI between those isolates collections.
Figure 3.
Bar chart of frequency distributions of the most prevalent CCs in 12 human datasets.
Only CCs of which a prevalence higher than 10% was found are plotted.
Figure 4.
PCA transformed vectors of CC frequencies in 12 human datasets.
The first, second and third PCA transformed dimensions explain together 73% of the total variability in the data. Weighted sums of the CC frequency distributions of the human isolates in the different datasets reported in Table 1 are plotted in the first two (upper graph) and in the second two (lower graph) dimensions of the PCA transformed space.
Figure 5.
Overall mean probability (%) and 95% confidence interval for human C. jejuni and C. coli infections to originate from chicken, cattle, pig, sheep, and the environment.
A. Baseline attribution results (see main text); B. Attribution results with Dutch chicken isolates replaced by chicken isolates from Scotland, the UK and Switzerland; C. Attribution results with Dutch chicken isolates replaced by chicken isolates from New Zealand, Finland and USA; D. Attribution results with Dutch, Scottish, English and Swiss chicken isolates as separate Campylobacter reservoirs.
Figure 6.
Statistics of the self-attributed proportions of 250 chicken isolates for reduced source datasets of size n (on x-axis).
Every reduced dataset is generated from the original dataset by randomly removing isolates from an original set of 150. The boxes indicate variability in the mean attributed proportions over the 10 different reduced datasets per model and per reduction factor. Indicated are the minimal, maximal and average means. The whiskers indicate the average 2.5% and 97.5% confidence limits over the different reduced datasets. The star-symbols represent the minimum 2.5% limit and the maximum 97.5% limit.