The University Animal Hospital offers highly qualified surgery and medicine and some patients are immunocompromised. Consequently, the usage of antimicrobials might be relatively high. Since antimicrobial usage has been described as a risk for equine MRSA colonisation [11
] effective infection prevention and control are essential in such settings. Success in short-term eradication by sectioning, improved infection control and repeated MRSA testing without using antimicrobials in two Canadian farms with MRSA-colonised horses has been reported [43
]. The main interventions in the present study were aimed at improved infection prevention and control, supported by environmental sampling.
The hospital was expanding at the time of the outbreak, with an increase of about 1000 patients per year. Increased patient flow with more horses housed in the hospital may have led to more indirect and perhaps also direct contacts between horses and thus a higher risk of transmission of infections. Increased crowding was also considered a risk of undesirable doping of horses by pharmaceutical residues in the environment. Therefore some precautions regarding handling of horses in the hospital had been applied, although not documented, before the MRSA outbreak.
There was a lag of over a month between surgery and diagnosis of the two first MRSA cases and thereby a delay in identification of the MRSA outbreak (Table ). The source of the outbreak was never found, since the first three horses infected had indirect cross-contacts within the hospital [37
]. The immediate suspension of elective surgery when an outbreak was suspected resulted in a temporary reduction in surgical patients. This gave space and time for necessary interventions in the surgery unit and probably prevented further spread of the MRSA strain. Emergency surgery continued, for animal welfare reasons. After elective surgery was suspended, only one horse, an emergency surgery patient, was infected by MRSA. The closure of Stable A, where Cases 1-5 were housed, was logical, as it could be expected to be highly contaminated with MRSA. This was confirmed by the first environmental sampling (Table ).
Representatives from different disciplines, the CV, a public health nurse and the SVA visited the UDS when an outbreak was suspected and started a process of change. Meeting face-to-face, walking around the hospital setting and discussing emerging issues from different angles in a mutual conversation opened the way for systematic work. Public health aspects had to be considered, as MRSA is a zoonosis [12
] which is notifiable for humans, and prevention of MRSA transmission between animals will reduce the risk for humans too.
The executive working group transferred some general infection prevention and control measures from human medicine, such as basic hygiene. Further interventions undertaken to curb the outbreak were based on this general knowledge and on observed practices and shortcomings, as well as on environmental sampling.
In order to improve the guidelines for handling and interventions in outbreak situations, knowledge of risk factors for acquisition and spread of MRSA in horses is needed. Horses colonised with MRSA on admission were more likely to suffer from clinical infection than non-colonised horses in one Canadian study [11
]. In a study of MRSA outbreaks in a Dutch equine hospital, 9.3% of horses tested positive on admission, and the same spa
-types were detected in infected horses [12
]. Such studies could speak in favour of screening on admission, but cannot effortlessly be transferred to low prevalence countries, as Sweden. Admission to neonatal intensive care was another identified risk factor for MRSA colonisation [42
]. About a year after the outbreak, a foal was admitted for neonatal care and MRSA was later detected in a pressure wound. The MRSA status of the foal and its mother on admission was not known. On readmission for infection in September 2009, the foal was isolated before culture and the strain did not cause further infections in other horses. Sampling of horses on admission to hospital had not been considered relevant prior to the outbreak described, as only one MRSA horse had been detected earlier in the country, a carrier in 2007 [18
]. At present there is still no general screening for MRSA in horses on admission to the UDS because of the low prevalence, although horses with discharging wounds are sampled.
Culture is time-consuming and therefore fast and reliable detection methods for screening on admission are needed. A real-time PCR validated for diagnostic purposes in humans has been evaluated, but proved unsuitable for use in horses [44
]. Screening of incoming risk patients might be recommended, e.g. horses with infected wounds, previously known MRSA infection/carriage, coming from high prevalence areas or hospitals with a known outbreak etc., but only if all these patients could be isolated prior to the sampling results. Studies in humans have shown that isolation of a patient only when a positive result is obtained does not reduce transmission to other patients if basic precautions are not consistently taken [45
Screening cultures are not wholly reliable and may give a false sense of security. There might even be a risk of staff handling screened horses testing negative with less care if horses are divided into MRSA-positive and negative groups, and thus the risk of contagious infection could increase for the negative group. Basic precautions and good cleaning routines must be implemented throughout the year and for every patient, irrespective of culture results. Isolation in single stalls should be implemented for risk patients such as those with discharging wounds.
Screening of asymptomatic contacts can be useful for epidemiological investigations and for interventions in an outbreak. Based on the severity of the disease and the risk to animal and human populations, SJV can initiate and pay for mandatory sampling. Contact tracing of discharged horses that had been operated on during the outbreak period led to the identification of one more infected horse, Case 5 [37
]. Nasal screening of the large numbers of asymptomatic horses that visited the hospital during the outbreak period would have required the owners' consent, as MRSA is not subject to mandatory tracing in animals and no resources were available for this. Therefore, there were gaps in the information and we do not know how the outbreak was maintained during the weeks between the first and second cluster of cases. Information about personnel is also lacking, as about two-thirds did not attend the sampling.
Environmental sampling was introduced in order to identify routes of indirect contact transmission. No standard or recommended method for environmental sampling of MRSA was available at the time, and the sampling and culture method was a combination of methods described in the literature and personal experience. The surfaces in equine hospitals are often rough and large areas have to be sampled and ordinary swabs or contact plates cover only limited areas. Here wiping of similar surfaces and pooling of samples was performed in order to cover as many items and as large an area as possible at a reasonable cost. The pre-impregnated cloths used have been tested in similar studies of environmental enterococci [38
]. The pre-impregnation neutralise possible residues of cleaning and disinfection agents. The sensitivity and specificity of the method is not known, as there is no 'gold standard'. Our culture method, with high-salt pre-enrichment broth and a second specific broth, was similar to the more sensitive method identified by v. Duijkeren et al. when comparing two methods [12
]. The majority of the positive samples in the UDS were found in connection with the outbreak and the number of positive samples decreased as expected after interventions were implemented (Table ). The second sampling (Table ) was carried out after major interventions had been initiated. Dividing the twice positive 21 doorknobs pooled sample into seven pools, when retesting 1 December 2008, were made to check if many doorknobs might be positive, but this time tested negative. Some time was then allowed to elapse for the interventions to bed down before the third sampling in June 2009, as the main purpose of repeated sampling was to monitor improvements.
The MRSA isolates found in the environment had the same PFGE pattern as the isolates from the outbreak horses described earlier [37
]. The isolate from last environmental sampling, in April 2010, showed a slight genetic change in PFGE, pulsotype A1, compared with the nine previous isolates (Figure ). This isolate was found in an ISO stall after a horse with the same pulsotype (A1) had been discharged. The isolate was still considered to have the same origin as the other nine. It is not uncommon for bacterial strains to change genetically over time and a one to two band difference in PFGE is considered to be closely related [41
In order to distinguish between human contamination and horse/human contamination, the sampling surfaces were divided into two categories, horse touch surfaces and hand touch surfaces (Table ). The results of the environmental sampling were used for educational purposes, to increase awareness of basic hygiene, and for the planning of interventions. Findings of the MRSA strain on hand touch surfaces indicated spread by humans (Table Figure ). The finding of MRSA on door knobs (Table Figure ) led to the installation of dispensers with alcohol hand-rub beside all doors, with the idea 'easy to find and do right'. Other ways of improving contact precautions (basic hygiene) were better clothing routines, the increased use of plastic aprons and gloves, and the replacement of disinfectant shoe baths/mats with disposable boots for isolation stalls.
The finding of MRSA in a sample from salt stones and cribs supported the decision to remove those and the water feeders and replace them with buckets. Buckets were easy to clean in the installed flusher-disinfector and they tested negative for MRSA on two occasions (Table ). An additional benefit was better water control for the patients. As MRSA was found in the pooled sample of the sling (tarpaulins) and traverse in the surgery unit, these were removed. Using disposable plastic sheets on the mobile operation tables was the unconventional solution introduced to keep the costs for disposable equipment at reasonable levels.
, including MRSA, can persist on dry inanimate surfaces for months [46
], and cleaned surfaces are rapidly recontaminated when infective patients are present. Therefore, the findings of MRSA where infected horses had been stalled were not surprising. Written disinfection and cleaning routines available to all were developed and replaced long-standing procedures for environmental cleaning and disinfection. Qualitative environmental sampling is not considered sensitive enough to validate cleaning routines and should not be used for that purpose [36
]. Despite this, in the present outbreak negative environmental cultures tended to give staff a false sense of security, as they tended to interpret these as successful eradication of MRSA.
Some items were more difficult to improve, e.g. floor coverings, which need to be rough for the safety of the horses but then become difficult to clean. The walls and floors in the surgery recovery rooms were worn out and considered a high risk for the spread of MRSA and were replaced with non-porous and easy to clean material. The remaining floors were not changed, since the outbreak coincided with construction plans for a new hospital building. Improvement of infection prevention and control in the daily clinical routines was gradually achieved but the conditions in the building were a limiting factor for cleaning.
The doubling in expenditure on gloves and disinfectants is a reasonable indication of improved standards of hygiene. On the other hand, MRSA was still found on surfaces that were not directly available to horses long after the last outbreak case of MRSA. Thus, compliance with basic hygiene and/or cleaning and disinfection routines has improved but is still not complete and requires continual follow-up to ensure maximum effect. Methods to measure compliance with hygiene routines need to be introduced in veterinary practice. Observational studies are a possibility to record improvements in hygiene routines. Quantifying the efficacy of the interventions would also be advantageous, but was not the aim of this study.
The interventions led to a reduction in MRSA-positive samples, and no new cases occurred for over a year, despite the extended presence of a colonised horse (Case 6). The case detected 14 months after the outbreak, a foal admitted for neonatal care in June 2009, had the same MRSA strain as the six outbreak horses in 2008 [37
]. The strain was also found on hand touch surfaces
in the surgery unit in June 2009. Whether this was the cause or the effect of the presence of an infected/colonised horse could not be determined, as we do not know whether the foal was carrying the strain on admission or became infected within the hospital. No other MRSA infections were detected in connection with this case. The case that occurred in March 2010 was also a single case without any other detected infections in the hospital. It is unlikely that pathogenic MRSA would remain undetected, as all infected wounds were cultured. The MRSA ST398, spa
-type t011, is common outside Sweden [12
]. A closely related strain caused an infection linked to another equine hospital in the area in 2009 [37
]. This indicates that the strain is present in the region outside the hospital studied here. We therefore consider Case 6 to be the last case in the outbreak and the other two later cases as sporadic.
The costs of an outbreak will differ between clinical settings and can be divided into different categories in order to have clear view of expenses: i) Costs directly linked to an outbreak, exemplified by suspension of elective surgery, extra working hours, isolation of horses, damage to reputation resulting in a future decrease in patients, bacteriological sampling, hire of a cleaning team and IC expert; ii) running costs to achieve general high quality, exemplified by increased expenditure on disposable gloves, other disposable items, soap, disinfectants, clothes, continuous education and revision of routines; and iii) depreciable investments, exemplified by renovation of recovery rooms, purchase of washer and flusher-disinfectors, changing from water feeders to buckets.
Distinction of costs into these categories was not done at the UDS. The calculation of total cost resulted in a rather high figure, but depreciable investments were included and it is debatable whether such investments should be fully included as costs due to an outbreak. Direct costs and running costs might also have been under- or over-estimated, as they are difficult to calculate exactly. If detailed distinctions had been made during the process, it would have been easier to analyse the cost elements justified for future infection control.
This outbreak and earlier incidents of MRSA in pets, mainly dogs, have resulted in growing interest in infection prevention and control policies by Swedish animal hospitals and clinics. MRSA has been notifiable since 1 January 2008 in animals. Guidelines for owners of MRSA-positive horses were published on the SVA website in July 2008. A Swedish Government Commission draft strategy from March 2011 against antimicrobial resistance and nosocomial infections drawn up by the National Board of Health and Welfare (Socialstyrelsen) includes veterinary medicine [47
]. Infection prevention and control has been introduced as a topic in the Veterinary Nursing programme at the Swedish University of Agricultural Sciences since 2009. This is an important step towards a more professional view of veterinary medicine infection prevention and control for the future.