抄録
The purpose of this study was multivariate analysis of relationship between use of antimicrobial agents and distribution of antimicrobial resistance in pig farms using the data of Japanese Veterinary Antimicrobial Resistance Monitoring System (JVARM). The antimicrobial resistance and husbandry data of 364 Escherichia coli isolates recovered from 251 pigs (one pig was sampled at each farm) between 2004 and 2007 were used in this study. In univariate analysis, the association between selection of antimicrobial resistance and the use of antibiotics and husbandry factors were tested. In multivariate analysis, Generalized Estimating Equations were used, choosing each antimicrobial resistance as response variable. As explanatory variables, the use of antibiotics, husbandry factors and interaction terms between them with a p-value less than 0.2 in univariate analysis were selected.
Ampicillin resistance was associated with the use of macrolides (MLS) (n=26, OR=2.7, p=0.002). Kanamycin resistance was associated with the use of MLS (n=14, OR=2.5, p=0.005), colistin (n=5, CL, OR=7.5, p=0.002) and polypeptide-based feed additives (n=20, OR=2.5, p=0.005). Oxytetracycline (OTC) resistance was associated with the use of aminoglycoside (n=25, OR=2.7, p=0.02), tetracycline (n=55, TC, OR=30.3, p=0.004) and MLS (n=45, OR=11.3, p=0.003), especially at younger age (TC (slope of logit=-0.49, p=0.05) and MLS (slope=-0.40, p=0.007)). Dihydrostreptomycin resistance was associated with the use of MLS (n=37, OR=2.4, p=0.003). Chloramphenicol resistance was associated with the use of TC (n=22, OR=2.4, p=0.007). Enrofloxacin (ER) resistance was associated with the use of fluoroquinolone (n=1, FQ, OR=17.1, p=0.02) and CL (n=1, OR=12.3, p=0.04). Nalidixic acid (NA) resistance was associated with the use of FQ (n=2, OR=8.3, p=0.03).
Cross-resistance was suggestive for OTC resistance due to the use of TC and ER and for NA resistance due to the use of FQ. Co-resistance was suggestive for several structurally-unrelated antimicrobials; however further analyses are required to estimate the casual relationships between use of drugs and resistance selection.
