*Contributed equally to this work.
Detection and Control of a Nosocomial Outbreak Caused by Salmonella Newport at a Large Animal Hospital
Article first published online: 15 MAR 2010
Copyright © 2010 by the American College of Veterinary Internal Medicine
Journal of Veterinary Internal Medicine
Volume 24, Issue 3, pages 606–616, May/June 2010
How to Cite
Steneroden, K.K., Van Metre, D.C., Jackson, C. and Morley, P.S. (2010), Detection and Control of a Nosocomial Outbreak Caused by Salmonella Newport at a Large Animal Hospital. Journal of Veterinary Internal Medicine, 24: 606–616. doi: 10.1111/j.1939-1676.2010.0484.x
- Issue published online: 7 MAY 2010
- Article first published online: 15 MAR 2010
- Submitted May 21, 2009; Revised December 11, 2009; Accepted January 28, 2010
- Bacterial species;
- Hospital surveillance;
- Infection control;
- Infectious diseases;
Background: Nosocomial salmonellosis is often assumed to occur because infection control and surveillance practices are inadequate, but published evidence is lacking to support the related contention that rigorous application of these practices can impact the severity of outbreaks.
Objective: Describe active surveillance, early recognition, and intensive mitigation efforts used in an effort to control an outbreak of nosocomial Salmonella enterica serotype Newport infections without hospital closure.
Animals: Large animals hospitalized at a referral hospital.
Methods: This prospective outbreak investigation was initiated when Salmonella Newport infections were detected among hospitalized animals by active surveillance. Data were analyzed to identify temporal and spatial patterns for epidemic spread of Salmonella in the hospital. Mitigation efforts were aggressively adjusted in response to surveillance data. Genetic relatedness of isolates was investigated by pulsed-field gel electrophoresis.
Results: Of 145 large animals sampled, 8 (5.6%) were infected with the Salmonella strain associated with this outbreak, and all but 1 shed Salmonella in the absence of or before the onset of disease. This strain was recovered from 14.2% (42/295) of environmental samples (ENV samples), indicating that widespread environmental contamination had occurred. Isolates of Salmonella Newport obtained from infected animals and the environment were genetically indistinguishable, confirming clonal dissemination.
Conclusions and Clinical Importance: Active surveillance allowed early detection of nosocomial Salmonella transmission and hospital contamination. Use of aggressive interventions was followed by cessation of transmission. Active surveillance can allow earlier recognition and mitigation compared with programs by only sampling of clinically affected animals.
Colorado State University Veterinary Diagnostic Laboratory
- ENV sample
James L. Voss Veterinary Teaching Hospital at CSU
pulsed-field gel electrophoresis
- S. Newport
Salmonella enterica serotype Newport
Nosocomial infections caused by Salmonella enterica are an important hazard for veterinary hospitals, especially large animal facilities.1–6S. enterica is the agent most commonly associated with outbreaks of nosocomial disease at veterinary teaching hospitals and is the most commonly cited reason for closure,1 and zoonotic infections are a common feature in published descriptions of outbreaks.7–9 The documented risks associated with nosocomial S. enterica infections have prompted the initiation of rigorous surveillance and infection control measures in many veterinary hospitals.1–4,10–18 Despite aggressive control efforts, severe outbreaks at times have forced veterinary hospitals to temporarily close as an extreme measure for stopping nosocomial transmission and to allow mitigation of environmental contamination. For example, most published accounts of nosocomial S. enterica outbreaks in veterinary hospitals have involved closure or restriction of admissions.2–4 This type of dramatic response is considered the most rigorous method available for controlling Salmonella epidemics and is sometimes considered the only response that can be used in extreme situations. However, hospital closure creates important financial losses, is perceived to adversely affect the reputation of hospitals, limits provision of care for animals, and interrupts education of veterinary students, interns, and residents. Therefore, it is useful to describe control and mitigation efforts that can be used without closing the hospital. The purpose of this report is to describe how active surveillance and rigorous mitigation allowed control of an outbreak of nosocomial Salmonella enterica serotype Newport (S. Newport) infections in a large referral hospital without resorting to closure.
Materials and Methods
Using data obtained through an active surveillance program, an outbreak of S. Newport infections was detected among hospitalized large animals in August and September of 2006 at the James L. Voss Veterinary Teaching Hospital at CSU (JLV-VTH). Data were analyzed to identify temporal and spatial relationships related to the spread of a single S. Newport strain among large animals and within the hospital environment. Data were summarized relative to the timing of standardized and augmented mitigation efforts that were employed during the outbreak. The genetic relatedness of the S. Newport isolates was investigated using pulsed-field gel electrophoresis (PFGE). Events in the outbreak were described as days elapsed since the admission of the 1st animal from which S. Newport was isolated, and the Epidemic Period was defined as the interval between the date of admission of the index case (July 28, 2006; Day 0 of the outbreak) and the date last positive sample was collected from the last affected animal (September 1, 2006; Day 35).
Medical records were reviewed for all large animals (new world camelids, goats, cattle, pigs, and horses) hospitalized during the Epidemic Period. Data retrieved from records included signalment, clinical signs, dates and locations where animals were housed, paths of animal movement in the hospital, diagnostic and surgical procedures performed, dates of sampling and culture results for fecal samples that were submitted for culture of S. enterica, clinical outcomes (survived and were discharged versus died), and necropsy findings of the horse that died.
In accordance with the previously established biosecurity protocols,19 fecal samples were routinely obtained from all hospitalized large animals at admission and then every Tuesday and Friday throughout hospitalization. Sampling frequency was increased to admission and Monday, Wednesday, and Friday in response to the outbreak. Samples were also regularly obtained from surfaces in the hospital to monitor for contamination with S. enterica. Environmental samples (ENV samples) obtained with electrostatic wipesa were collected monthly at 63 predetermined sites located throughout the small and large animal areas of the JLV-VTH as previously described.13,19 ENV samples were also routinely obtained from stalls after they were vacated by large animals with known or suspected high risk of S. enterica infection. A minimum of 12 hours was allowed to elapse after cleaning and disinfection before obtaining ENV samples.
Collection of ENV samples occurred as previously scheduled on Day 17 of the outbreak. Subsequently, as mitigation efforts proceeded, multiple sets of ENV samples were obtained from targeted areas to determine the extent of environmental contamination and efficacy of mitigation efforts. These areas included the 63 sites routinely sampled in addition to additional areas that were sampled to clarify the extent of contamination.
Culture of S. enterica
All culture and susceptibility testing was performed at the Colorado State University Veterinary Diagnostic Laboratory (CSU-VDL).13,19 The CSU-VDL is an American Association of Veterinary Laboratory Diagnosticians accredited laboratory and operates with the requisite quality control standards. The earliest that fecal cultures could be interpreted as being positive was approximately 2 days after submission, and samples were never concluded to be culture negative until at least 3 days after submission. The earliest that culture of ENV samples could be interpreted as being positive was 5 days after submission and cultures were not concluded to be culture negative until at least 7 days after submission. All isolates were tested to determine serogroup as described,13 and to evaluate antimicrobial susceptibility using agar diffusion methods.20 These results were usually available within 2 days of presumptive identification of S. enterica isolates.
All isolates were also sent to a reference laboratory for serotyping.b Because results of serotyping were not available for a month or longer after submission, during the outbreak period S. enterica isolates were assumed to have spread through clonal dissemination if they had the same serogroup and antimicrobial susceptibility pattern. After serotyping, isolates determined to be S. Newport that had the same antimicrobial susceptibility pattern were designated as the ‘Outbreak Strain.’
Genetic relatedness of S. Newport isolates was investigated at a reference laboratory after the outbreak by PFGE.c Representative isolates included in this comparison were purposefully selected from each of the animals found to be shedding the Outbreak Strain, as well as 4 isolates recovered from ENV sampling sites distributed throughout the facility. Additionally, S. Newport isolates recovered from samples submitted to the CSU-VDL from locations in Colorado (not the JLV-VTH) were compared with isolates recovered during the outbreak. Specifically, a S. Newport isolate that was the strain most recently recovered from a horse, cow, and dog before the onset of the outbreak was purposefully selected for comparison. The 24-hour S. enterica PFGE procedure was performed as described by the National Molecular Subtyping Network for Foodborne Disease Surveillance (Pulse Net).21
Biosecurity Standard Operating Procedures at the JLV-VTH
Infection control and biosecurity policies and procedures were described in detail and were available online to all personnel, and all students, staff and house officers received formal orientation before working with animals.19 It was a stated expectation that all personnel working in the JLV-VTH know and comply with all aspects of these policies and procedures.
Briefly, all hospitalized cattle, goats, new world camelids, and pigs were housed in the Agricultural Animal Ward (Fig 1). All animals hospitalized in the Agricultural Animal Ward were housed in an individual stall that had solid sidewalls, and slatted metal gates in the front and rear of the stall. A footbath containing a peroxymonosulfate disinfectantd was placed at the entry gate to each stall and at the entry and exit points of the Agricultural Animal Ward. All personnel wore clean coveralls and rubber overboots while handling animals. Animals housed in the Agricultural Animal Ward that were known or suspected to be shedding S. enterica were maintained in this ward with additional biosecurity precautions.
Horses known or suspected of shedding S. enterica or other contagious agents were hospitalized in the Equine Isolation building. Horses admitted for other gastrointestinal conditions were hospitalized in the equine colic ward. Other hospitalized horses were housed in the general equine inpatient areas, but were segregated by service (eg, medicine, surgery). All horses hospitalized in the Equine Ward were housed in individual stalls with solid side and rear walls and a slatted entry gate. Footmatse containing a peroxymonosulfate disinfectant were placed at the entry and exit points for the aisles leading to different sections of the Equine Ward, and at the gate for all colic recovery stalls and at the entry of all isolation stalls. Standard cleaning and disinfection procedures included removal of gross contamination, scrubbing surfaces with detergent and sodium hypochlorite (1:32), rinsing, and application of quaternary ammonium disinfectant (1:256).f,19 Additional measures were used when cleaning and disinfecting stalls and other areas associated with managing high risk cases (eg, colic aisle and isolation). These included an additional application of sodium hypochlorite solution (1 : 12), and high-pressure, high-temperature rinsing as a final step. Additionally, a 4 times normal concentration (ie, 4%) solution of peroxymonosulfate disinfectant was periodically applied as a mist to all surfaces in the Large Animal Hospital.14
Summary of S. enterica Shedding and Environmental Contamination
There was a clear difference in the overall rate of recovery and the recovery of S. Newport isolates when comparing the Epidemic Period to time periods preceding this outbreak (Table 1). Before this outbreak, S. Newport had not been recovered from active surveillance of animals admitted to the JLV-VTH since December 2004, and had not been recovered from an ENV sample since June 2005. In the 7 months before the Epidemic Period, only 1.1% of (6/545) of large animals and 2.5% (13/517) ENV samples were positive for S. enterica, and none of these isolates were serotype Newport. Specifically, in the 3 preceding monthly ENV sample sets, 0 of 63 were positive in May, 1 of 63 in June, and 2 of 63 in July.
|Time Period||S. enterica Serotype||Percent Salmonella-Positive (Number/Total)|
|Environmental Samples||Fecal Samples from Hospitalized Large Animals||Hospitalized Large Animals|
|January 1 to July 28||Newport||0%||0%||0%|
|Other||2.5 (13/517)||0.7 (7/1008)||1.1 (6/545)|
|July 29 to September 1||Newport||19.7 (37/188)||7.3 (18/245)||5.6 (8/145)|
|(Epidemic Period, Days 1–35)||Other||4.8 (9/188)||5.7 (14/245)||6.2 (9/145)|
|September 2 to September 29||Newport||4.3 (6/139)||0%||0%|
|(Days 36–63)||Other||11.5 (16/139)||7.7 (14/182)||14.2 (13/91)|
|September 30 to December 31||Newport||0%||0.2 (1/489)||0.4 (1/269)|
|Other||13.3 (33/249)||5.1 (25/489)||9.8 (24/269)|
In contrast, between Day 0 and Day 35, S. enterica was recovered from 24.5% (46/188) of ENV samples and 12.4% (18/145) of hospitalized large animals (Table 1). A majority of these isolates were subsequently determined to be serotype Newport (70.5%; 55/78). Because of requisite delays in obtaining serotype classification from the reference laboratory, definitive information regarding serotypes was not available until after the end of the outbreak. It should be noted that the rate of recovery was also higher for other serotypes of S. enterica (ie, not Newport) during the Epidemic Period for both fecal and ENV samples, which complicated real-time interpretation of culture results. Twenty-three other hospitalized animals (19 cattle, 3 horses, and 1 pig) were shedding other S. enterica serotypes, including Schwarzengrund, Typhimurium, Infantis, Mbandaka, Montevideo, Oranienburg, Meleagridis, Muenster, and Cerro. Other strains of recovered from 34 ENV samples included serotypes Meleagridis, Typhimurium, Mbandaka, Montevideo, Give, and Putten.
Early in the outbreak, it was difficult to interpret with certainty the importance of recovering S. Newport isolates as these isolates matched by antimicrobial resistance pattern, those that had were commonly found in the referral population. For example, of the 499 S. enterica isolates recovered from active surveillance of hospitalized animals and the hospital environment for approximately 4 years before the Epidemic Period, between July 1, 2002 and July 28, 2006, 148 (29.7%) were serotype Newport. Among these Newport isolates, 88% (131/148) had an antimicrobial susceptibility pattern identical to the Outbreak Strain.
S. enterica isolates recovered from the initial case and all other isolates considered to be associated with this outbreak were serotype Newport (serogroup C2), and all had identical antimicrobial susceptibility patterns. Isolates were resistant to amoxicillin-clavulanate, ampicillin, ceftiofur, cephalothin, chloramphenicol, streptomycin, sulfonamide, and tetracycline and were susceptible to amikacin, gentamicin, enrofloxacin, and trimethoprim-sulfamethoxazole. There were not epidemiological ties among other S. enterica isolates (those that were not S. Newport) recovered during the Epidemic Period. The few isolates that were similar to each other were recovered from cows originating from the same dairies, but recovery from admission samples and the fact that these cows were from the same premises suggested that infections occurred before admission.
Through the Epidemic Period, a total of 8 hospitalized large animals were found to be shedding the Outbreak Strain of S. enterica (4 alpacas, 2 horses, 1 goat, and 1 cow). Four of these animals showed clinical signs compatible with salmonellosis before discharge, but active surveillance detected shedding of the outbreak strain in 7 of the 8 animals in the absence of clinical signs or before the onset of disease. No S. enterica infections were identified among hospital personnel or clients in association with this outbreak. However, information regarding human infection and illness was collected passively and it is possible that subclinical infections occurred or that clinical infections were not reported or were not attributed to S. enterica infections.
The 1st culture-positive fecal sample was collected from the Index Case on Day 11 (August 8, 2006), and the last positive sample was collected from the last affected animal on Day 35 (Fig 2). Among all 34 fecal samples collected from the infected animals during the outbreak period, 42% (14/34) were positive for S. Newport. Additionally, after detection of fecal shedding in the Index Case, the Outbreak Strain was recovered from 52 ENV samples that were obtained from widely distributed locations in the VTH (Fig 1). Environmental contamination was first reported from the laboratory on Day 21 of the outbreak, and results of the last positive ENV samples were reported after the Epidemic Period on Day 69.
The 1st isolate of the Outbreak Strain was cultured from a 2-year-old female alpaca (Alpaca A, Figs 2 and 3) as part of routine hospital surveillance. This animal was admitted to the Agricultural Animal Ward on July 28, 2006 (Day 0) for evaluation of ileus and related GI signs. Fecal samples obtained on Days 1 and 3 were culture negative for Salmonella, but the Outbreak Strain was recovered from the 3rd sample (results reported on Day 13) and also from 3 subsequent samples. Alpaca A developed diarrhea on Day 11.
In the 7 subsequent cases, recovery of the Outbreak Strain was first reported from the laboratory on Days 18, 19, 23, 24 (2 animals), 27, and 35 (Fig 2). All animals originated from different premises, except for Alpacas C and D, which were dam and cria. Alpaca B was an 8-year-old female alpaca that was admitted for evaluation of pneumonia. It had 1 negative culture before developing diarrhea on Day 14, and the Outbreak Strain was recovered from 3 subsequent fecal samples. Alpaca C was a 9-day-old alpaca cria that was admitted for evaluation of generalized weakness, which was later determined to be associated with a vertebral body abscess. The cria's 3-year-old dam (Alpaca D) was admitted at the same time and was housed with the cria. Fecal samples from both animals were culture negative on Day 14, before development of diarrhea in Alpaca C on Day 18 and subsequent detection of S. enterica shedding. Alpaca D remained subclinical throughout its hospitalization. Horse A was a 10-year-old Quarter Horse gelding that was originally admitted for evaluation of lameness on Day 0; 2 fecal samples collected during this initial hospitalization were culture negative. This horse was discharged on Day 7 after initial evaluations and then readmitted on Day 18 for arthroscopic surgery. A fecal sample collected at readmission was culture negative, but the horse became febrile and neutropenic on Day 21 and was moved to isolation. The laboratory reported recovery of the Outbreak Strain from a fecal sample on Day 23. This horse developed enterocolitis, diarrhea, and septicemia and subsequently died on Day 25. At necropsy, severe necrotizing, suppurative typhlocolitis was evident, and the Outbreak Strain was recovered from cultures of small intestine, liver, lung, and spleen (Fig 3). Recovery of the Outbreak Strain was reported from Goat A (17-year-old female) and Cow A (5-year-old female) on Days 24 and 27, respectively. Neither of these animals showed signs associated with S. enterica infection while hospitalized. Additionally, recovery of the Outbreak Strain was reported on Day 35 from a fecal sample collected from Horse B (6-year-old female). This horse was admitted on Day 21 for orthopedic surgery, fecal samples collected on Days 26 and 28 were culture negative, and it did not show signs associated with S. enterica infection while hospitalized.
Environmental Contamination with S. enterica
During and following the Epidemic Period (Days 17–63), a total of 327 ENV samples were collected, including 32 samples collected from stalls in the main hospital (Tables 1 and 2). Excluding stall samples, the Outbreak Strain was isolated from 14.2% (42/295) of ENV samples and other unrelated strains of S. enterica were isolated from 7.5% (22/295; Table 1). The Outbreak Strain was most commonly recovered from areas where infected animals were housed (Agricultural Animal Ward and Equine Isolation), but was also recovered from other widely dispersed areas throughout the facility (Equine Ward and other areas of the JLV-VTH, Table 1 and Fig 1).
|Area Sampled||Sample No.||Recovery of Salmonella enterica|
|Percent Positive for Outbreak Strain (No.)||Percent Positive for Other Strains (No.)|
|Agricultural Animal ward||79||26.6 (21)||15.2 (12)|
|Equine ward||115||4.3 (5)||4.3 (5)|
|Equine Isolation||33||33.3 (11)||0.0 (0)|
|Other JLV-VTH areas||57||3.5 (2)||7.0 (4)|
|CSU veterinary diagnostic laboratory||8||25.0 (2)||0.0 (0)|
|Parking Lot||3||33.3 (1)||33.3 (1)|
|Large Animal stalls (excluding isolation)||32||3.1 (1)||9.4 (3)|
|Total||327||13.1 (43)||7.6 (25)|
Among ENV samples obtained during the outbreak, 56.3% (166/295) were obtained from floor surfaces within the JLV-VTH, 18.3% (55/295) from hand-contact surfaces, and 22.0% (65/295) from composite samples of both hand-contact and floor surfaces. Additionally, 3.1% (9/295) of ENV samples were obtained from other types of surfaces that are not typically monitored (rubber boots, parking lot, etc.). For all ENV samples, S. enterica was isolated from 14.5% (8/56) of hand-contact surfaces, 22.9% (38/166) of floor surfaces, and 24.6% (16/65) of composite (hand-contact and floor) surfaces. Among the ENV samples collected from stalls that housed high-risk animals after cleaning, only 3.1% (1/32) were culture positive for the Outbreak Strain, and 9.4% (3/32) were positive for unrelated strains of S. enterica.
Laboratory results indicating that animals were shedding the Outbreak Strain became available for Alpacas A, B, and C on Days 14, 17, and 20, respectively (Fig 2). In accordance with standard biosecurity operating procedures,19 increased precautions were initiated in the Agricultural Animal Ward on Day 14 when results indicated that a hospitalized animal was shedding S. enterica. Considering that Alpacas A and B did not have known epidemiological ties before admission and considering the similarity of the S. enterica isolates, when these results became available on Day 17, it was strongly suspected that shedding in Alpaca B was associated with nosocomial transmission.
That day, personnel in the Agricultural Animal and Equine Wards were notified of the problem and instructed to implement heightened biosecurity procedures, including mandatory wearing of exam gloves when touching animals hospitalized in the Agricultural Animal Ward. Additionally, nursing staff and students removed materials, and thoroughly cleaned and disinfected surfaces in personnel areas (records room, offices, etc.) in the Agricultural Animal Ward (Fig 1) with quaternary ammonium disinfectant. Concerns were also relayed to personnel that this multidrug-resistant, serogroup C2 isolate appeared similar to highly contagious and pathogenic S. Newport strains that had caused several serious outbreaks in veterinary hospitals and other animal populations in the United States (Aceto, University of Pennsylvania, personal communication, 2008).22,23
On Day 20, the Outbreak Strain was isolated from a 3rd animal (Alpaca C). Precautions enacted in response to this additional, apparent nosocomial infection included movement of Alpacas B, C, and D to the Equine Isolation building and immediate discharge of all remaining animals from the Agricultural Animal Ward. New admissions to the Agricultural Animal Ward were voluntarily limited to emergency situations. Otherwise, animals were cared for in the field by ambulatory clinicians from the JLV-VTH. Clients were informed of the reason for these precautions (ie, potential nosocomial S. enterica transmission) whenever they called to schedule inpatient services.
From Day 20 until the conclusion of the outbreak, all personnel working in the large animal wards of the JLV-VTH were provided detailed information to increase overall awareness and elicit careful adherence to hygiene and biosecurity precautions. This included information regarding affected animals such as culture results, clinical history, clinical condition, housing and previous movement within the JLV-VTH, biosecurity precautions, and plans for further actions such as intensive cleaning. Less detailed communications were periodically sent to all students, faculty, and staff.
On Day 21, preliminary culture results from ENV samples routinely collected as part of ongoing monthly surveillance on Day 17 became available, indicating that 22.2% (14/63) of samples were culture positive for the Outbreak Strain, and unrelated strains of S. enterica were recovered from 3 other samples. The Outbreak Strain was recovered primarily from the Agricultural Animal Ward (10 sites, Fig 1), but also from floor surfaces in the CSU-VDL (2 sites) and from the Small Animal Hospital (2 sites). While additional cleaning and disinfection was already underway within the Agricultural Animal Ward, plans were revised to further intensify hygiene and biosecurity efforts in all areas of the large animal wards. Surveillance in hospitalized animals was also intensified: beginning on Day 21, samples were collected from all large animals at the time of admission and every Monday, Wednesday, and Friday through the remainder of the outbreak. From Days 20 to 24, each stall in the Agricultural Animal Ward was cleaned and disinfected by standard methods.21 On Day 24, peroxymonosulfate disinfectant solution (4 times normal strength) was applied by mist application.
Despite implementation of intensive control measures, on Day 24, results of fecal samples obtained on Day 21 indicated that 2 additional hospitalized animals were shedding the Outbreak Strain. These were Goat A and Horse A, the latter being the 1st infected animal housed outside of the Agricultural Animal Ward. In response, a full set of 63 ENV samples was collected on Days 24 and 25 to evaluate the effectiveness of decontamination of areas that had been cleaned and disinfected and to determine if contamination had spread to other areas in the JLV-VTH. Additionally, biosecurity precautions were increased throughout the equine general inpatient area of the JLV-VTH. Specifically, rubber overboots were required to be worn by all personnel when working in this area and footbaths containing normal strength (1%) peroxymonosulfate were placed in front of each stall that housed an animal. In addition, personnel contact with animals was limited whenever possible, the need for excellent hand hygiene was emphasized to all personnel, movement of large animals within the hospital was restricted to that which was absolutely necessary (eg, walking of large animals for therapy or exercise was discontinued). Discharge of animals colonized with S. enterica was considered a priority (when their medical conditions allowed) in order to reduce the potential for further spread. By Day 26, no animals known to be shedding the Outbreak Strain remained in the hospital. All owners were alerted to the shedding status of their animals before discharge and were counseled in depth about precautions that needed to be taken to minimize the likelihood of transmission of S. enterica to people or other animals at their home premises.
On Days 28 and 29, culture results became available for ENV samples collected on Days 24 and 25. Despite the mitigation efforts that had been initiated, these results indicated that environmental contamination was still widely distributed in the large animal hospital: 13 of 63 (20.1%) ENV surveillance samples were culture positive for the Outbreak Strain and additional strains of S. enterica were recovered from 4 other samples. The Outbreak Strain was again primarily recovered from the Agricultural Animal Ward (9 sites), but also from surfaces in equine inpatient areas (3 sites) and from the Equine Isolation building (1 site; Fig 1). Based upon these results, admissions of horses for elective procedures were rescheduled whenever possible, and plans were developed to repeat the cleaning and disinfection of areas where contamination had been detected. Again, clients were informed of the reason for these precautions (ie, the potential nosocomial S. enterica transmission) whenever they called to schedule inpatient services.
Surfaces in Equine and Agricultural Animal Wards were cleaned and disinfected by standard methods.19 Afterward, on Day 32, a 4 times regular strength solution of peroxymonosulfate disinfectant was applied as a mist14 in the breezeway, and all areas in the Agricultural Animal Ward. Plans were also made to intensively clean and disinfect all areas of the Equine Ward in a 2-phase effort. To enable the 1st phase, the few hospitalized horses remaining in the south end of the equine inpatient area were moved to stalls in the northwest section of the equine inpatient ward on Day 33. After standard cleaning and disinfection, these areas were disinfected on Day 35 by mist application of 4 times regular strength peroxymonosulfate disinfectant solution.
A revised set of 51 ENV samples was collected on Day 33, and results that became available on Day 38 indicated that 7 of 51 (13.7%) of cultures yielded the Outbreak Strain (3 from Equine Isolation facilities, 2 from the north end of the equine inpatient facilities, and 2 from the Agricultural Animal Ward), while 2 others yielded other strains of S. enterica. Additionally, results became available on Day 35 indicating that Horse B was shedding the Outbreak Strain. This horse was immediately moved to the Equine Isolation facility. It remained in isolation until negative results were obtained from a series of 5 fecal cultures obtained on Days 37 through 41, after which it was discharged. Horses remaining in the hospital were moved to the south end of the equine general inpatient ward on Day 37 (Fig 1); horses admitted for colic were segregated from other hospitalized horses, and standard operating procedures were maintained. All surfaces in the north end of the equine general inpatient areas, as well as areas in the Equine Isolation facility and the Agricultural Animal Ward that were found to be culture positive were cleaned and disinfected by standard methods.19 Subsequently, on Day 40, the northern half of the equine inpatient facilities, the Agricultural Animal Ward, and the Equine Isolation facilities were disinfected by mist application of 4 times regular strength peroxymonosulfate disinfectant solution.
A set of 53 ENV samples were collected on Day 45, and another set of 49 on Day 52. In both sets, samples obtained in 1 stall in the Equine Isolation facility were positive. Additionally, this stall was positive when sampled on Day 63. After each of these positive results were obtained, the isolation facility was intensively cleaned and disinfected by standard cleaning procedures as well as mist application of 4 times regular strength peroxymonosulfate disinfectant. Samples taken after the 3rd time this facility was cleaned and disinfected were finally reported as being negative on Day 75.
After the last recovery of the Outbreak Strain on Day 63, S. Newport was not recovered from hospitalized animals or ENV samples for several months despite active surveillance (Table 1). S. Newport was not recovered again from a large animal until December 19, 2006, and was not recovered from an ENV sample until July 24, 2007. Other serotypes of S. enterica recovered from the hospitalized large animals during this subsequent period included Give, Havana, Infantis, Mbandaka, Meleagridis, Montevideo, Muenster, Oranienburg, Schwarzengrund, and Typhimurium.
Genetic Comparisons of Isolates
On Day 32, representative S. enterica isolates that had been recovered during the outbreak were sent to a reference laboratory for genetic analysis. Results of these analyses became available on Day 46, indicating that isolates previously suspected of being spread through nosocomial dissemination were indistinguishable based upon analysis of PFGE results (Fig 3). In contrast, outside isolates that were analyzed for comparison purposes were distinguishable from the Outbreak Strain, but the genetic fingerprints showed varying degrees of similarity. The outside isolates that had the same antimicrobial susceptibility pattern as the Outbreak Strain (isolates P and Q) had greater similarity (97% and 95% similarity with the Outbreak Strain, respectively) than did the isolate that had a very different susceptibility pattern (isolate R) which showed only 77% homology. In comparison with genetic patterns in the USDA's VetNet database which contains PFGE typing results for 1,270 S. Newport isolates from throughout the United States, isolates with genotypes identical to the Outbreak Strain (pattern JJPX01.0170) represent only 1.5% of S. Newport isolates. In contrast, isolate Q was representative of the most common genotype in the VetNet database (pattern JJPX01.0023, 16.3% of S. Newport isolates), while the genotype for isolate P (pattern JJPX01.0048) represented 2.5% of S. Newport isolates and the genotype for isolate R (pattern JJPX01.0338) represented 0.08% of S. Newport isolates in the VetNet database.
During this outbreak, a strain of S. Newport spread quickly from the Index Case to 7 additional hospitalized animals, causing clinical disease in 4 and death in 1 horse. Additionally, in spite of intensive control measures, the Outbreak Strain was spread widely throughout the JLV-VTH and persisted even after rigorous cleaning and disinfection. Highly infectious and contagious isolates, widespread environmental contamination, and environmental persistence are common features among major S. enterica outbreaks that have previously forced closure of large referral hospitals, including the JLV-VTH (Aceto, University of Pennsylvania, personal communication, 2008).2–4 Active surveillance for S. enterica facilitated early detection of the Outbreak Strain and we believe influenced our ability to minimize the consequence of this outbreak. Of the 8 animals found to be shedding the Outbreak Strain, all but 1 shed Salmonella in the absence of clinical signs or before the onset of disease. Therefore, had it been our practice to only sample animals with clinical signs compatible with salmonellosis (eg, diarrhea, fever, leukopenia), only 4 of 8 animals known to be infected would have been detected, and detection would have been delayed by several days, potentially allowing further spread among hospitalized animals and in the environment. As such, we believe that closure of our hospital was an imminent threat that was avoided primarily because of aggressive surveillance and mitigation strategies, albeit with difficulty.
The most probable source for initial dissemination of the Outbreak Strain was the Index Case (Alpaca A). It is likely that large numbers of bacteria were shed by this animal in association with its GI disease, and contamination was spread in the Agricultural Animal Ward by personnel carrying out their required duties and also through the movement of the alpaca. Before S. enterica was detected in its feces, this animal was walked repeatedly inside and outside the Agricultural Animal Ward to promote resumption of normal GI motility, an activity that likely facilitated spread of the S. Newport to other areas. Culture results from ENV samples obtained on Day 17 (results reported on Day 21) provide an indication of this spread; the Outbreak Strain was recovered from 14 of 63 ENV samples, including 2 samples from the Small Animal Hospital. Integration of culture data from hospitalized animals and the environment across time and location enabled us to better define the scope of the outbreak and initiate targeted cleaning and disinfection measures, as well as initiate more rigorous animal handling protocols.
It is impossible to know exactly when hospitalized animals were exposed to the Outbreak Strain given the uncertainty imposed by variable incubation periods, variable shedding, and imperfect diagnostic tests. In retrospect, it is possible that a majority of the animals were infected during a very short, intensive exposure period. Six of 7 subsequent cases could have been infected in as few as 4 days (Days 18–22), but it is also possible that animals were exposed for 33 days (Days 0–33) before all nosocomial infections occurred. All animals had been hospitalized for several days before obtaining fecal samples from which the Outbreak Strain was recovered, with the exception of Cow A. The positive fecal sample obtained from this cow was collected per rectum with a gloved hand within 8 hours of admission and was placed directly into a sterile container. It seems unlikely that the sample was contaminated from the environment. While the genetic fingerprint of this isolate was identical to the Outbreak Strain, it is possible that this cow was infected before admission. The outside isolates with identical antibiograms that were evaluated were genetically similar but not identical (Fig 3). Very little published literature is available which critically evaluates the incubation period of S. enterica in large animal species. However, outbreaks of food-borne salmonellosis are often intensively investigated by public health officials, and the period between exposure and onset of clinical disease is reportedly dose dependent and can be as short as 8–12 hours.24,25 Thus, if the cow were highly susceptible and exposed to a high dose of a highly infectious strain of S. enterica, it seems possible that it could have been exposed in the hospital and shed the outbreak strain.
As demonstrated in this outbreak, it is critical to differentiate strains of S. enterica in order to identify dissemination of a single S. enterica clone within a population comprised of animals from multiple sources. Unfortunately, the requisite lag time associated with obtaining culture results and phenotypic information (serogroup and antibiogram) complicates identifying and managing spread of S. enterica within veterinary hospitals. Two to 5 days was required after sample submission to obtain culture results and serogroup information, and an additional 1–3 days was needed to obtain antimicrobial susceptibility profiles. To further complicate matters, differentiation of strains from the same serogroup by antibiograms is tentative at best, but it often requires a month or longer to obtain more definitive serotype information from reference laboratories in the United States. While PFGE can also be used to determine the genetic relatedness of multiple isolates, this type of analysis is expensive and is not widely available, as was the situation for the CSU-VDL during this outbreak. Polymerase chain reaction (PCR) is sometimes used to analyze fecal samples and ENV samples to identify S. enterica. However, a positive result from this assay indicates the presence of genetic material from S. enterica, but does not allow other phenotype or genotype characterization that is required to identify clonal dissemination. In light of the fact that strains of S. enterica other than the outbreak strain were detected through active surveillance during this outbreak, it is possible that the scope of the outbreak (number of affected animals and number of ENV samples) could have been overestimated, had a nonstrain specific, PCR-based surveillance system been in use. Additionally, positive PCR results could theoretically be obtained when samples are contaminated with dead organisms or even DNA fragments. This could be particularly troublesome when analyzing ENV samples collected as follow-up to decontamination measures.
The need to differentiate among strains of S. enterica in this type of animal population is illustrated well by surveillance data obtained during the outbreak. The Outbreak Strain was detected in 4.2% (18/427) of fecal samples, while 6.7% (29/427) of fecal samples yielded unrelated strains (Table 1). Similarly, while cultures of 14.7% (43/292) of ENV samples yielded the Outbreak Strain, unrelated S. enterica strains were recovered from an additional 8.2% (24/292) of ENV samples. While biosecurity measures were not different for animals that were found to be shedding unrelated S. enterica strains, the implications and the responses are substantially different than when infections were thought to have resulted from nosocomial transmission. Most notably, evidence of nosocomial transmission of S. enterica is more likely to initiate hospital-wide control measures than is detection of multiple unrelated Salmonella strains in hospitalized animals.
Despite the importance of this outbreak, the Large Animal Hospital at the JLV-VTH remained open during this outbreak, although measures were taken to reduce admissions through provision of ambulatory service care and rescheduling of elective procedures. In contrast, during the S. enterica outbreak which occurred at the JLV-VTH in 1996, 56 animals were infected, 3 deaths occurred and the hospital was closed for a 3-month period.2 In 2001, a similar number of animals were infected (7) and died (1), but the hospital was closed for a 1-month period to allow extensive decontamination efforts.14 While closure might have been considered a more conservative and therefore more appropriate response to this outbreak, the benefits of such a closure must be measured against the adverse impacts on the health of animals denied admission, the financial status and reputation of the hospital, and educational opportunities provided by the veterinary teaching hospital. Two factors were considered pivotal at the time that this decision was considered during the outbreak. First, results of cultures of fecal and ENV samples, interpreted in the light of long-term experience with active surveillance at the JLV-VTH, led us to conclude that we had detected clonal dissemination of S. enterica in hospitalized animals and the environment soon after it was initiated. Second, in the time since our previous challenge of this magnitude, we had identified and evaluated methods for rapid, wide-scale environmental decontamination of veterinary facilities.14
Movement of hospital personnel and animals within the facility likely contributed to the dissemination of environmental contamination in this outbreak. Among the ENV samples, those obtained from floor surfaces were more frequently positive than those obtained from hand surfaces, suggesting that foot traffic of humans or animals, or both, contributed to environmental dissemination, in spite of the presence of multiple disinfectant footbaths throughout the large animal facility. Detection of the Outbreak Strain in ENV samples obtained from hand-contact surfaces indicates that personnel contributed directly to environmental dissemination, despite a strict hand hygiene policy that had been in place before and during the outbreak. Oral exposure of hospitalized animals and personnel would seemingly be more likely to occur when hand surfaces are contaminated.
Considering the rigorous mitigation efforts involving increased communication and awareness, increased restrictions for contact with animals and for movements within the hospital, repeated ENV and animal sampling, repeated cleaning and disinfection of various areas of the JLV-VTH, and increased use of barrier nursing/personal protective equipment, the hospital administrators in consultation with the personnel managing infection control in the hospital judged that the mitigation efforts were sufficient to preclude the need for closure. It is possible that if the hospital had been closed early in the course of this outbreak, some of the infections could have been avoided. However, it is not possible to accurately judge the impact of this measure retrospectively. Clearly, decisions regarding management of a nosocomial outbreak can be controversial and it might not be possible to fully determine a best course of action at the time that decisions are made, or even after the event has resolved. Regardless, it should be noted that throughout this episode, the JLV-VTH adhered to its policy of open disclosure of information regarding infection control and zoonotic hazards. While it might be tempting to guard this information closely when faced with this type of event, veterinary hospitals have an ethical and legal obligation to disclose information regarding infectious disease risks to hospital personnel and to clients.
aSwiffer, Proctor & Gamble, Cincinnati, OH
bU.S. Department of Agriculture-National Veterinary Services Laboratories, Ames, IA
cU.S. Department of Agriculture—Agriculture Research Service—Bacterial Epidemiology and Antimicrobial Resistance Research Unit, Athens, GA
dVirkon-S, Suffolk, UK
eFootbath Mats, Gemplers, Madison, WI
f456N, Ecolab Inc, St Paul, MN
The authors thank personnel at the JLV-VTH for their infection control efforts, and Denise Bolte, Dr Doreene Hyatt, and the CSU-Veterinary Diagnostic Laboratory for their diagnostic assistance.
- 5Chlorine dioxide gas decontamination of large animal hospital intensive and neonatal care units. Applied Biosafety 2006;11:144–154., , , et al.
- 9Centers for Disease Control and Prevention. Outbreaks of multidrug-resistant Salmonella typhimurium associated with veterinary facilities—Idaho, Minnesota, and Washington—August 24, 1999. Morb Mortal Wkly Rep 2001;50:701–704.
- 17Biosecurity. In: SellonDC, LongMT, eds. Equine Infectious Diseases. St Louis, MO: Saunders-Elsevier; 2007:528–539., , ,
- 18Biosecurity and infection control for large animal practices. In: SmithBP, ed. Large Animal Internal Medicine, 4th ed. New York: Elsevier; 2008:1524–1550.,
- 19Colorado State University. Biosecurity standard operating procedures, James L. Voss Veterinary Teaching Hospital. Available at http://csuvets.colostate.edu/biosecurity/biosecurity_sop.pdf. Accessed January 22, 2010.
- 20NCCLS. Performance Standards for Antimicrobial Disk and Dilution Susceptibility Tests for Bacteria Isolated from Animals, Approved Standard, 2nd ed. Wayne, PA: NCCLS; 2002.
- 24US Food and Drug Administration. Bad Bug Book - Salmonella spp. Available: http://www.fda.gov/Food/FoodSafety/FoodborneIllness/FoodborneIllnessFoodbornePathogensNaturalToxins/BadBugBook/ucm069966.htm. Accessed January 22, 2010.
- 25Centers for Disease Control and Prevention. Salmonellosis. Available at http://www.cdc.gov/nczved/dfbmd/disease_listing/salmonellosis_gi.html. Accessed January 22, 2010.