Longitudinal serological responses to Salmonella enterica of growing pigs in a subclinically infected herd.

A longitudinal survey was conducted in France in a subclinically Salmonella-infected farrow-to-finish pig farm to describe the time-course of the serological response to Salmonella enterica in growing pigs. We used three batches of sows and their corresponding litters (n = 31 litters). Among these, 256 pigs randomly selected and individually identified were followed from the first week of age until slaughter. Serial individual blood samples were submitted to indirect Salmonella-ELISA testing. Salmonella shedding was investigated by bacteriological testing of faecal material regularly collected from the sows and pigs and by environment swabs taken from the pens. Caecal contamination was checked at the slaughterhouse. Information about litter composition (filiation), location of the pigs in post-weaning and fattening pens, sanitary events, sex and body weights was recorded. 11.6% of the pigs shed S. enterica; 52% of pigs seroconverted before slaughter. The age-related variation of the natural logarithm of calibrated optical densities (COD) of pigs was described with two linear mixed models. From 8 to 61 days of age, the decrease in COD with age was fitted with a model including random effects of the animal and the dam on the intercept and slope, a batch random effect on the intercept and an individual birth-weight fixed effect on the intercept. The dam random effect was explained by the parity of the sow, Salmonella shedding by the sow during the farrowing phase and the value of the optical density of colostrum collected at parturition. A second model fitting the increase in COD from 61 days of age until slaughter included the random effect on intercept of the batch and the random effects on slope and intercept of the animal, the dam and the pen in which the followed animals were located during the fattening phase and the environmental contamination as fixed effect. In this second model, no relation was found between individual slaughter-bacteriological results and increasing COD values. Considering seroconversion time between 61 days of age and slaughter, survival function were constructed using the Cox proportional-hazards model. Both methods suggested that seroconversions generally occurred during the last third of the fattening phase (from 140 days of age to slaughter), while shedding was observed during the first half of the fattening period. The fitted models suggest the existence of clusters (such as pen and litter of origin).