The temporal pattern and relationship of Campylobacter prevalence in broiler slaughter batches and human campylobacteriosis cases in Sweden 2009–2019

Thermophilic Campylobacter species are the most common cause of bacterial gastroenteritis worldwide, and handling and consumption of broiler meat is considered a major foodborne transmission route. Both the incidence of campylobacteriosis and the prevalence of Campylobacter in broilers show seasonality but the impact of this association and broiler prevalence on human incidence is not clear. To explore this relationship we applied two approaches for analyzing time series data using different time resolutions (weekly, bi-weekly or monthly data) of human campylobacteriosis cases and prevalence of Campylobacter spp. in broiler slaughter batches in Sweden between 2009 and 2019. The decomposition of time series into seasonal (S), long-term trend (T) and residual components (STL model) showed a close overlap in seasonal patterns in terms of timing and the proportional change of peaks from normalized yearly levels. Starting 2016, when a large outbreak was reported, there was significant overlap in the trend components as well. The trend component of human cases prior to the outbreak corresponded to a linear increase of 6.5 % cases annually. In comparison, the estimated annual increase in broiler consumption was 2.7 %. An additive approach for time-series counts incorporating seasonal and epidemic (cases are a function of previous cases) components found a positive association between human cases and broiler prevalence with an optimal lag of 2 weeks, 1 bi-week, or 0 months. Considering the estimated time between slaughter and consumption, incubation time, and the time between on-set of disease and testing, a 2-week lag may be consistent with transmission via handling and consumption of fresh broiler meat. The best model included broiler prevalence as a factor in the epidemic model component, not in the seasonal component. The outcomes in terms of best model, optimal lags and significance of parameters, using weekly, bi-weekly or monthly data were, in general, in agreement but varied with data resolution when only a subset of the time series, not including any known broiler associated outbreaks, was analyzed. The optimal resolution based on the available data and conditions of the present analysis appeared to be weekly or bi-weekly data. Results suggest that broiler prevalence with a 2 week lag period can explain part of the human cases but has a smaller explanatory impact during the part of the study period not including the large known outbreaks. There is no simp