This study shows that antimicrobial resistance is widespread in enteric E. coli
from healthy as well as from diarrheic preweaned dairy calves. Streptomycin, sulphonamide, tetracycline or ampicillin were the most prevalent resistance traits and isolates resistant to all these antimicrobials were common. The findings are in agreement with studies in preweaned dairy calves from other countries [17
]. Also quite similar results were obtained from Swedish calves sampled at post-mortem [23
]. In such calves a high occurrence of resistance can be anticipated since a large proportion of the animals are probably treated with antimicrobials. An equally high prevalence of resistant E. coli
in the untreated calves of the present study is therefore remarkable.
As discussed by Call et al. [24
] the epidemiology of resistant E. coli
in calves is multifactorial, complex and e.g. influenced by co-selection due to linkage of resistance genes. But widespread resistance is fundamentally a consequence of historical and current use of antimicrobials and associations between use of antimicrobials and resistance in enteric E. coli
of calves have been documented [19
]. However, in the present study resistance is not a direct sequel to antimicrobial use since no calf was treated prior to sampling. Antimicrobial use in calves is still not uncommon in Sweden and Ortman & Svensson [29
] showed that in dairy herds about one third of diarrhoeic calves (1-90 days) were treated with trimethoprim/sulfa, enrofloxacin or other antimicrobials. Also in the present study NCD was treated with antimicrobials in several herds but there was no statistical association between such routines and resistance. Although the total use of antimicrobials in the herds is unknown, the absence of association indicates that the high prevalence of resistance is not solely an effect of a direct selection pressure by use of antimicrobials to the calves.
Importance of other aspects than antimicrobial use on prevalence of resistant E. coli
in preweaned calves is indicated by several studies [17
] and was recently reviewed by Call et al. [24
]. One proposed factor is a linkage between resistance genes and genes conferring selective advantage to colonize the intestinal lumen of calves. Walk et al. [28
] hypothesized that, regardless of use of antimicrobials, tetracycline resistance in E. coli
is co-selected in calves by an unknown "beneficial mutation". Likewise, Khachatryan et al. [21
] showed that E. coli
with the resistance phenotype streptomycin - sulfonamide - tetracycline have a selective advantage to colonize the intestine of calves given a dietary milk supplement also in absence of antimicrobials. Notably, in the present study about one third of the resistant isolates and two thirds of the multiresistant isolates had streptomycin - sulfonamide - tetracycline in their phenotype.
A selective advantage of resistant strains due to dietary differences is in agreement with the age related occurrence of resistant E. coli
documented elsewhere [18
]. The importance of age is evident also on comparison of data from the present study to previous data from older cattle in Sweden, in which resistant E. coli
are rare [23
Another factor of possible importance for resistance in preweaned calves is feeding milk from cows treated with antimicrobials or feeding colostrum from cows treated in the dry period [17
]. It has been proposed that antimicrobial residues of such milk could select for resistance in the enteric flora of calves. Although there are few studies to support the assumptions, the risk of a "hidden" selection pressure in dairy calves warrants further studies of the issue.
In the present study, resistance was more common in E. coli
from calves with diarrhea than in isolates from healthy calves. Also multivariate analysis showed that resistance to tetracycline in E. coli
was associated with diarrhea in calves. A higher occurrence of resistance in E. coli
in calves from herds experiencing problems with NCD was shown also by Gunn et al. [34
]. A plausible reason for this is that antimicrobials are used more often in herds with a high disease incidence, as indicated in the present study by the more common routine of treating NCD with antimicrobials in CD+ herds than in CD- herds. However, a higher prevalence of resistant E. coli
in calves with diarrhoea could also be due to linkage between virulence genes and resistance genes as proposed by Martinez & Baquero [35
]. Such linkage was demonstrated in E. coli
from pigs with diarrhoea [36
] and could imply co-selection of virulence genes by use of antimicrobials and conversely maintenance of resistance in populations of pathogenic bacteria as proposed by Boerlin et al. [36
]. In the present study phenotypic resistance to one or more antimicrobials was associated with presence of the virulence genes espP, irp
in E. coli
. However, there was no association to single antimicrobials or resistance phenotypes. Moreover, none of these virulence factors were associated with diarrhea and the findings should be interpreted with caution. Likewise, Suojala et al. [38
] found an association between the virulence factor iucD
and resistance to streptomycin, ampicillin, sulphametoxazole and trimethoprim in E. coli
from dairy cows with mastitis, but they found no association between the virulence factor and patogenicity.
Attention has been drawn by veterinary practitioners to the possibility that E. coli
is a more prevalent cause of NCD than demonstrated in previous Swedish studies, restricted to E. coli
with virulence factor F5. This suggestion is supported by international studies showing a higher prevalence of E. coli
F5 than in the Swedish studies [39
]. This virulence gene was not found in any isolate in this study and 21 out of 40 of the investigated genes were not detected or rarely detected. Although the number of isolates tested was small, this contradicts a common occurrence of enteropathogenic E. coli
in NCD. A possible exception is E. coli
with the virulence gene terZ
which was associated with diarrhea. However, further studies are needed to clarify clinical relevance of these virulence genes in NCD.
The etiology of diarrhea in this study was not clarified, since other infectious agents than E. coli
were not searched for. The impact of Salmonella
and BVDV, however, was most likely none, since the prevalence of Salmonella
spp is low in Swedish cattle herds [43
] and all the participating herds were declared BVD-free.
In the present study there was no association between diversity of enteric E. coli
and calf diarrhea. A decrease in the homogeneity of the faecal coliform flora has been shown in suckling pigs with diarrhea [44
]. Therefore, biochemical fingerprinting was performed to increase the probability of selecting isolates of the pathogenic strain and thereby the probability of detecting virulence genes. The results in the present study agree with the findings of Acha et al. [39
] and could indicate that in calves, there is no predominance of single clones even in diarrhea caused by enteropathogenic E. coli
. Still, in the present study it more likely indicates that diarrhea caused by E. coli
was uncommon since there, with one exception, was no association between occurrence of virulence genes and diarrhea.
In most studies of antimicrobial susceptibility in intestinal bacteria, isolates are randomly selected from a culture of intestinal content. But in the present study, isolates of the most common phenotype in each sample was selected for testing. This did not influence the overall prevalence of resistance in the group of calves sampled, but in individual calves it occasionally had a profound effect on the outcome of the susceptibility test. Accordingly, to guide the choice of antimicrobial therapy it can be misleading to test E. coli with undefined clinical relevance. Moreover, this emphasizes the advantage of identifying pathogenic isolates for susceptibility testing, possibly by detection of virulence factors.
Sampled calves deviated slightly from the given inclusion criteria, i e 0-3 (CD-) and 5-10 (CD+) of the 20 most recently born calves had been diarrheic, as compared to 0-2 (CD- criterion) and > 6 (CD+ criterion). Other sampling errors were a mismatch in age at sampling between CD+ and CD- calves, and an extended age interval for CD+ and CD- calves (1-135 days, mean 21.9) compared to the criteria (four weeks). Nevertheless, there were clear differences in clinical signs between CD- and CD+ calves and CD- and CD+ herds regarding mortality, occurrence of diarrhea, general appearance and respiratory signs. One contributing reason for this could be that CD+ herds were larger than CD- herds. Larger herds were more likely to have calf group feeder in the present study, and there might be a larger number of neonatal calves in the herd during the same time period which increases the risk for infectious enteric diseases. The association between diarrhea in calves and large herd size agrees with the findings in a recent Swedish study on calf morbidity and mortality in herds with different size (C Sandgren personal communication). The calf mortality in this study was higher than previously reported in Swedish herds [45
] which also could reflect a negative impact of large herd size on calf health. Calf rearing practices have changed concurrently with herd size.