periods: one from 22th March 2010 to 22th October 2010 (7 months
defined as the “warm period”) and the other including the rest of the
months (5 months defined as the “cold period”).
As the prevalence was expected to be lower during the warm than
cold period, the sampling scheme was as follows (Table 1). During the
warm period, 40 pigs in each of 60 batches were sampled; this total of
2400 pigs sampled was calculated to be sufficient to detect a minimal
pig prevalence (percentage of positive pig) of 2% with 25% relative
precision and 95% confidence for this period. During the cold period,
20 pigs in each of 36 batches were sampled; the total of 720 pigs
sampled was sufficient to detect a minimal pig prevalence of 15% with
20% relative precision and 95% confidence.
A database was created by compiling all the slaughter dates for the
16 selected slaughterhouses, that were coded A01 to A16. According
to the period, two or three batches were sampled each week. The
batches to be sampled were determined by random selection from a
list of slaughter dates and times from the database. Pigs to be sampled
within a batchwere chosen randomly using a systematic randomprocedure
based on a randominterval determined according to the size of the
batch (Dohoo et al., 2003).
The batches selected for sampling were sampled directly on
the slaughter line at the end of dressing, immediately before carcass
refrigeration. The soft palate tonsils of the required number of pigs
were swabbed with Cervibrush swabs (Cellpath, United Kingdom),
with the same technique as used in the preliminary study (Fondrevez
et al., 2010).