This study was carried out in 6 identical nursery units (A1 to A3 and B1 to B3) on the experimental pig farm at the Institute for Agricultural and Fisheries Research (ILVO, Merelbeke, Belgium). Each unit consists of 8 pens of 1.8 m2. Piglets were moved to these units immediately after weaning (4 weeks of age) and stayed there for 6 weeks. Each pen housed 6 piglets. Pen flooring was a synthetic grid, under which a board slopes towards a centrally-located slurry pit. Units A1 to A3 were monitored during 2 successive vacancy periods in February and April 2015 and units B1 to B3 during 1 vacancy period in March 2015. After pig removal, units were soaked with water. The day after, units were cleaned with hot water (80 °C), then disinfected with 1 % (v/v) MS Megades (Schippers, Bladel, The Netherlands) on the same day. The disinfection product consists of glutaraldehyde and quaternary ammonium compounds. After cleaning and disinfection, the pen remained vacant for 10 days. During this vacancy period, temperature and relative humidity (RH) were monitored hourly using thermo-hygrometers (Ilog EI-HS-D-32-L, ESCORT data logging systems). Three random pens per unit were sampled before disinfection and at 1, 4, 7 and 10 days after disinfection. Per sampling moment, 135 samples were taken, for a total of 675 samples.
Sponge swab samples (3 M, SSL100, St. Paul, MN, USA), pre-moistened with 10 mL Ringers solution (Oxoid, BR0052G, Basingstroke, Hampshire, England), were taken at 5 locations per pen: floor, concrete wall, synthetic wall, drinking nipples and feeding trough. Sampling of 3 pens per unit resulted in triplicates per type of location or 15 swab samples per unit at each time point. To neutralise the residual action of the disinfectants on the microbiological growth, 10 mL Dey Engley neutralising broth (Sigma Aldrich, Fluka, D3435, St-Louis, MO, USA) was used to pre-moisten the sponge swab samples that were used on day 1 after disinfection. A surface of 625 cm2 (A4 paper format) was sampled whenever possible. Because the surface of the drinking nipples was smaller than 625 cm2, 2 drinking nipples per pen were sampled. Samples were transported to the lab under refrigeration and were processed immediately. For all measured pathogens, selected relevant parameters and enumeration or detection techniques were based on Luyckx et al. . Swab samples were first diluted with 30 mL of Buffered Peptone Water (BPW, Oxoid, CM0509) and then homogenised by placing them in a Masticator (IUL instruments, S.A., Barcelona, Spain). Prior to plating, swab samples were further diluted in dilution series in saline peptone water (Bio Trading, K110B009AA, Mijdrecht, The Netherlands) to produce countable results on the selected agar media: Plate Count Agar (Oxoid, CM0325) for total aerobic flora and Slanetz and Bartley (Oxoid, CM0377) for Enterococcus spp. (lower enumeration limit 30 CFU/625 cm2). Plate Count Agar and Slanetz and Bartley plates were incubated at 30 and 37 °C during 72 and 48 h, respectively. A 10 mL BPW fraction was also transferred to a Stomacher® bag and mixed with 10 mL double concentrated Mueller Hinton Broth (Oxoid, CM0405) and 13 % (w/v) sodium chloride (Merck, 1.06404.500, Darmstadt, Germany). After overnight incubation of this solution at 37 °C, 100 μl was plated on chromID® MRSA SMART (MRSM, bioMérieux, Marcy l’Etoile, France) for the detection of MRSA. ChromID® MRSA SMART were incubated at 37 °C for 24–48 h. The remaining BPW fraction (original sample) was also incubated overnight at 37 °C for additional analyses: for detection of E. coli and faecal coliforms, 10 μl of the enrichment broth was plated onto Rapid E. coli medium (Biorad, 356–4024, Marnes-la-Coquettes, France) and incubated for 24 h at 44 °C.
The distribution of the log-transformed enumerations of total aerobic flora and Enterococcus spp. was analysed via graphs (Q-Q plot and histogram). The log-transformed enumerations of total aerobic flora followed a normal distribution. A linear regression model was conducted to evaluate the effect of a vacancy period and location on the log-transformed total aerobic flora enumerations (dependent variable). To assess the effect of predictor variables (vacancy period and location) on the non-normally distributed outcome variables, variables describing the enumeration and detection of the different bacteria (Enterococcus spp., E. coli, faecal coliforms and MRSA) were transformed into binary variables (absent or below the detection limit = 0, present = 1). Subsequently a logistic regression analysis was carried out. Temperature and RH were added as covariates in both models. Variable “unit” was included as a random effect in both models to correct for measurements within one unit.
Post-hoc comparison was performed with a Tukey-Kramer test. P-values ≤ 0.05 were considered as significant. All statistical analyses were carried out using Statistical Analysis System software (SAS®, version 9.4, SAS Institute Inc., Cary, NC, USA).