This study confirmed the first nosocomial outbreak of MRSA in a Swedish equine hospital. This was also the first MRSA infections notified in horses in Sweden, although one nasal carrier of spa-type t011 had been found in a screening survey of horses about six month earlier, in December 2007 . As the screening was anonymous, it could not be determined whether there were connections between that finding and the subsequent outbreak. The outbreak strain belonged to the MRSA variant ST398, spa-type t011 [38, 39]. This variant has been associated with livestock in Europe [12, 19] and with horses [15, 17, 18, 28, 29], but had only been found in horses and a few humans  in Sweden.
The ST398, spa-type t011, isolates from Swedish horses were indistinguishable or very closely related according to PFGE, which corroborated the close relationship between them. The use of two restriction enzymes offered good discrimination between the isolates and revealed minor genetic change over time, from pulsotype A into pulsotype A1 in Hospital 1 (2010, Horse IX) and pulsotype A2 in Hospital 2 (2009, Horse VII). The pulsotypes A1 and A2 were isolated more than a year after pulsotype A and genetic events in a strain leading to changes in restriction site are not surprising, according to Tenover et al.  and indicated one circulating strain in the region.
It was not possible to find a plausible source of the outbreak since the index case could not be defined, as the first three horses infected had indirect cross-contacts within the hospital. Moreover, horses were not screened for MRSA on arrival at the equine hospital, probably because MRSA had not yet been identified as a risk in Sweden. Horses regularly travel to competitions and for breeding and change owners over regional and national borders. The MRSA in this study could well have originated from connections to livestock or horses in countries with a higher MRSA prevalence than Sweden. As MRSA had not been found in livestock in Sweden at that time, it seemed less likely that Swedish livestock would be the source [39, 40]. The infection might even have derived from human carriers, such as people handling the horses, as professionals working with animals that carry MRSA ST398 have been reported to be colonised with this type [2, 15, 20–23]. The infected horse found during telephone tracing of the owners of horses that had undergone surgery during the outbreak period in Hospital 1 might have been overlooked otherwise, as the owner expressed no worry about secretions from the wound. This indicates a need for post-operative monitoring of horses in order to track down surgical complications, but also to communicate general information about signs of resistant bacteria to veterinary staff, horse owners and other people working with horses. The findings also point out a possible bias in estimated numbers of hospital-acquired infections based upon infections diagnosed solely within the hospital or clinic.
A different spa-type, t064, was introduced to Sweden almost two years after the outbreak in Hospital 2 (Table 1 Horse X). This spa-type has been reported in horses in other countries [12, 15, 49, 50]. In Hospital 2 another horse (Table 1 Horse VIII) suffered from MRSA pulsotype A2, which was a minor genetic change of pulsotype A from the Hospital 1 outbreak. The two cases in Hospital 2 occurred approximately six months apart and was apparently not an outbreak, but the incident initiated a need for revision of the infection control procedures within the hospital. Such revision and improvement had already started at Hospital 1.
The majority of the MRSA infections identified in this study were SSI, confirming previous reports, although other infections do occur [8–10, 12, 14, 15, 17, 28, 51].
The zoonotic potential of MRSA means that infected and carrier horses could pose a risk to hospital staff and horse owners. Seven human cases of spa-type t011 were notified between 2006 and 2010 in Sweden , and the findings of spa-type t011 in people could be the result of animal to human transmission. It would be interesting to compare Swedish spa-type t011 MRSA isolates from people and horses by PFGE. In a Finnish study, isolates of spa-type t011 from people without horse contact showed a different PFGE pattern than t011 isolates from a veterinarian and horses in contact during a MRSA outbreak in an equine hospital .
The isolates tested here were resistant to all antimicrobials approved for equine use in Sweden and the only treatment options were drugs with a risk of inducing colitis in horses, e.g. erythromycin  and one for local use, fusidic acid. Spontaneous clearing of the infections without antimicrobial treatment was an encouraging observation, as there was no approved drug available in the studied cases and antimicrobial resistance is expanding and antimicrobials are considered a limited resource [54, 55]. Wound care without antimicrobial treatment is a topic that needs to be further explored within veterinary practices, as well as in research.
The great variation in time from surgery to MRSA detection (15-52 days) was not acceptable, as delayed detection might influence the numbers of horses infected in an outbreak, as well as posing a hazard to staff. Some of the delay could be explained by negative cultures on the first sampling occasion and at least one unaware owner, the telephoned-traced case. Other reasons for delay were empirical antimicrobial treatment before sampling (Horse X) and delayed sampling (Horse I). Lack of awareness or neglect by veterinarians and others are hypothetical reasons for delays. The findings showed that infections should be taken seriously and would benefit from qualified sampling before antimicrobial use, although negative culture is a risk, as demonstrated in two of the cases in this study. The time between bacteriological sampling and laboratory confirmation of MRSA was also long in some cases. Culture for MRSA is a balance between speed of result, sensitivity, specificity and cost. A suitable sampling technique is vital for faster culture. Pre-washing of the top layer of an infected wound with sterile saline, water or likewise to remove decomposition flora might help to decrease the delay that arises from analysing mixed flora, especially if MRSA selective agar plates are not used and the growth of decomposition flora requires recultivation. However, there could also be delays because of lack of awareness or dismissal of MRSA by laboratories, since MRSA had not been detected previously in Sweden. Accurate diagnosis is important for all concerned; patient, owner, veterinary staff, etc. Verification of MRSA isolates by PCR is essential, since a false diagnosis of MRSA is as devastating as a true positive. Faster methods would be preferable to obtain a quick yes or no answer, but culture will still be required to get isolates for antimicrobial susceptibility testing and genetic and epidemiology studies. Real-time PCR as used for humans has been evaluated for detecting MRSA in horses, but proved unsatisfactory .
The MRSA outbreak raised an alarm for equine veterinarians and hospital staff in Sweden and increased their awareness about the need for improved infection control for their own safety and that of the horses. In addition, during the outbreak, the topic was reported and discussed frequently and thoroughly in the media, making the general public aware of the presence of MRSA bacteria in the equine population.