• Open Access

Outbreak of Salmonellosis Caused by Salmonella enterica Serovar Newport MDR-AmpC in a Large Animal Veterinary Teaching Hospital

Authors

  • B.L. Dallap Schaer,

    1. Department of Clinical Studies, University of Pennsylvania School of Veterinary Medicine, Kennett Square, PA
    Search for more papers by this author
    • *Contributed equally to the production of the manuscript.

  • H. Aceto,

    1. Department of Clinical Studies, University of Pennsylvania School of Veterinary Medicine, Kennett Square, PA
    Search for more papers by this author
    • *Contributed equally to the production of the manuscript.

  • S.C. Rankin

    1. Department of Pathobiology, New Bolton Center, University of Pennsylvania School of Veterinary Medicine, Kennett Square, PA
    Search for more papers by this author

  • This work was presented in part at the 11th International Symposium on Veterinary Epidemiology and Economics, Cairns, Australia, August 2006.

Corresponding author: B. L. Dallap Schaer, Department of Clinical Studies, New Bolton Center, University of Pennsylvania School of Veterinary Medicine, 382 West Street Road, Kennett Square, PA 19348; e-mail: bldallap@vet.upenn.edu.

Abstract

Background: Nosocomial salmonellosis is an important problem for large animal veterinary teaching hospitals (VTHs).

Objective: To describe failure of an Infection Control Program (ICP) that resulted in an outbreak of salmonellosis caused by Salmonella Newport multidrug resistant (MDR)-AmpC at a large animal VTH.

Animals: Sixty-one animals identified with the outbreak strain of Salmonella.

Methods: Retrospective study: Data collected included signalment, presenting complaint, duration of hospitalization, discharge status, and financial information. Phenotypic and genotypic characterization was performed on Salmonella isolates.

Results: The outbreak occurred despite an existing ICP; the ICP was reviewed and weaknesses identified. Routine patient surveillance was not performed before or during the outbreak; fecal sampling was triggered only by a patient algorithm based on clinical signs. Sixty-one animals were infected with the outbreak strain of S. Newport, and the majority were horses (n = 54). Case fatality rate was 36.1%. S. Newport isolates demonstrated high genetic similarity (Dice ≥ 0.96), and all had the MDR-AmpC phenotype. Environmental persistence of the organism necessitated complete hospital closure, extensive decontamination, and remediation of the facility. A paradigm shift in the relevance of biosecurity in a VTH and the establishment of a stringent ICP were integral components of successful hospital reopening.

Conclusions and Clinical Importance: An ineffective ICP resulted in a nosocomial outbreak caused by a MDR S. Newport in a VTH. Closure of a VTH affected all missions of the institution and had substantial financial impact (US$4.12 million).

Abbreviations:
GI

gastrointestinal

ICP

Infection Control Program

ICU

intensive care unit

MAC

MacConkey II agar

MDR

multidrug resistant

MOD SURV

methods for environmental surveillance for Salmonella implemented late April 2004

NBC

New Bolton Center

NICU

neonatal intensive care unit

ORIG SURV

methods for environmental surveillance for Salmonella in place from 1984 to late April 2004

PADLS

Pennsylvania Animal Diagnostic Laboratory System

PAS

perinatal asphyxia syndrome

PCR

polymerase chain reaction

PFGE

pulsed-field gel electrophoresis

SRC

Salmonella Reference Center at New Bolton Center

VTH

Veterinary Teaching Hospital

XLD

xylose-lysine-deoxycholate agar

Several risk factors for fecal shedding of Salmonella have been identified in hospitalized large animals.1–10 Risk factors range from clinical signs and treatment1–10 related to the presenting complaint to uncontrollable environmental factors such as distance travelled to the hospital2,4,6,7 and daily ambient temperature.5 Reports describing risk factors have been derived from studies of sporadic shedding in hospitalized animals as well as investigations of nosocomial outbreaks. Patient groups at increased risk for shedding Salmonella include horses with diarrhea,1–4,6 horses with colic,1,2,4,5,10 and those presenting with fever.3 Nosocomial salmonellosis in veterinary teaching hospitals (VTHs) has resulted in hospital closure,11–14 leading to disruption of clinical service and suspension of clinical rotations. Closure of a VTH affects patient welfare, incurs financial losses, and hampers the teaching mission. In a recently published survey of biosecurity experts at 38 VTHs, 31 (82%) hospitals reported the occurrence of a nosocomial outbreak in the 5 years before the survey interview.15Salmonella enterica was most commonly identified as the cause of outbreaks (20/31, 65%), and was the agent most commonly cited as the reason for restricting admissions (17/22, 77%).15 Because horses exist as subpopulations connected through extensive population mixing,16 VTHs can act as nodes in the contact network. The role of the referral hospital in disease amplification and dissemination of infectious agents is an important consideration.17

The nosocomial outbreak reported here was caused by Salmonella Newport multidrug resistant (MDR)-AmpC, which is distinguished by presence of the plasmid-mediated ampC gene (blaCMY218,19) that encodes resistance to extended-spectrum cephalosporins. The outbreak occurred despite having an Infection Control Program (ICP) in place, was characterized by a high infection and case fatality rate, and had substantial financial impact. To the authors' knowledge, analysis of an existing ICP, the role its failure played in an extensive outbreak, and the financial impact of closure and remediation have not been reported previously. The purposes of this investigation were to retrospectively confirm that a perceived increase in salmonellosis was a nosocomial outbreak caused by a single strain of Salmonella, identify the failures in infection control measures that allowed the outbreak to occur, and describe the impact of this adverse event on the large animal VTH.

Materials and Methods

Existing ICP and Patient Surveillance

An ICP had been in place for approximately 20 years before the investigation, and was administered by an Infectious Disease Committee. Patient protocol consisted of separate housing for animals admitted with diarrhea or strong clinical evidence of zoonotic or nosocomial disease, patient surveillance triggered only by clinical signs of diarrhea for >12 hours, and regular environmental surveillance. Diarrhea combined with fever (≥102°F) and leukopenia (<4,000 cells/μL) resulted in the patient being moved to isolation, as did the persistence of diarrhea for >36 hours. Patients were not segregated by risk for nosocomial disease, although most food animals were housed in a designated barn. Horses admitted with colic were routinely triaged out of the intensive care unit/neonatal intensive care unit (ICU/NICU) facility into a barn that also housed outpatients or horses admitted for short elective procedures. There was no definition of what constituted a nosocomial infection.

Original Environmental Surveillance (ORIG SURV)

Before the outbreak, environmental surveillance (ORIG SURV) was performed as outlined in the existing ICP. Five individual samples were collected every 2 weeks from 1 of 5 predetermined high-traffic locations in the hospital (selected barn locations, postmortem room, patient admission area, surgery areas). Results were to be reviewed by the Quality Assurance Committee, which met quarterly, or the Infectious Disease Committee, which convened as needed.

Sample collection involved swabbing (1 stroke) a dry, sterile gauze sponge over a dry surface, and placing it in a sterile sample collection bag.a The sample was processed as described below for fecal culture, with selenite brothb enrichment and differential plating techniques. Colonies identified as Salmonella were selected, and antimicrobial susceptibility testing, serogrouping, serotyping, and pulsed-field gel electrophoresis (PFGE) analysis were performed. Samples were stored in the Salmonella Reference Center (SRC) for further analysis if necessary.

Modified Environmental Surveillance (MOD SURV)

After telephone consultation with a biosecurity specialist (Dr Paul Morley, Colorado State University School of Veterinary Medicine) approximately 2 weeks before closure, environmental sample collection20 and culture methods were changed. Sample collection with a commercially available electrostatic wipec and sweeperd passed over a broad surface within a sample area was implemented, and areas were divided into hand and foot samples. Samples were processed by a modified International Standards Organization (ISO) culture procedure.21 This is a multistage process involving pre-enrichment, enrichment, and differential plating designed to select for Salmonella. Presumptive Salmonella colonies were identified, antimicrobial susceptibilities were determined, and isolates were serogrouped and serotyped as described below.

In the early stages of the investigation, financial and administrative restrictions limited the number of samples collected and analyzed. In an effort to determine the extent of hospital contamination during investigation of the outbreak, 120 samples were submitted from areas of human traffic and animal facilities within the Widener Hospital (April 27, 2004). After closure, an additional 222 samples were collected to determine the extent of contamination throughout the facility. To ensure appropriate decontamination and remediation throughout the clean-up process, >1,100 samples were submitted and processed with the MOD SURV technique.

Animals

Medical records of all patients admitted from July 1, 2003 (date of admission of the retrospectively identified index case) through May 10, 2004 (date of hospital closure) in which a positive identification of Salmonella was made on at least 1 sample (fecal, blood, or postmortem intestinal contents) were reviewed. Signalment, presenting complaint, days of hospitalization, hospital location, billing information, and outcome were recorded. All patients that had ≥1 positive culture for Salmonella were investigated retrospectively. Patients were classified as affected by nosocomial salmonellosis if Newport MDR-AmpC was identified on culture. After confirmation of a nosocomial outbreak, all teaching and research animals on the campus were screened for Salmonella. Pooled samples were submitted from nonequine species housed in numbers >10.

Culture Techniques

Fecal samples and gastrointestinal (GI) contents from postmortem examination were processed with overnight selenite enrichment and differential subculture to Xylose-Lysine-Deoxycholate (XLD)e and MacConkey II (MAC)f agar plates. Organisms presumptively identified as Salmonella were serogrouped with commercially available antisera.g Antimicrobial susceptibility testing was performed with an automated system and commercially available plates,h and breakpoints were determined by the manufacturer according to Clinical and Laboratory Standards Institute (formerly National Committee for Clinical Laboratory Standards, NCCLS) guidelines.22–24 Serotypingi was performed by standard World Health Organization approved techniques.25

During the study period, blood cultures were routinely collected from foals admitted to the NICU with suspected sepsis or perinatal asphyxia syndrome (PAS). Blood was collected aseptically and the appropriate volume placed into commercially available blood culture vials.j Blood culture vials were processed as described by the manufacturer; no additional enrichment steps specifically selective for Salmonella were used. Organisms identified as Salmonella were handled as described above.

PFGE and PCR Assays

Pulsed-field profiles from PFGE were used to confirm genetic homogeneity of strains isolated during the outbreak. PFGE was performed according to the standard Centers for Disease Control and Prevention (CDC) PulseNet USA protocol.26 DNA fragments were visualized on an ultraviolet transilluminator and photographed with a digital camera system.k Digital image files were stored and further analyzed, including creation of dendrograms by softwarel designed to compare PFGE profiles. Individual profile numbers were assigned to each isolate with at least 1 band difference in the PFGE pattern.

Detection of blaCMY and blaTEM genes was performed in the SRC by PCR as described previously.18 Detection of blaSHV was performed at the CDC by a modification of a previously reported method.27 Additional details of the methods for total genomic DNA extraction and amplification are described elsewhere.28,29

Statistical Analyses

Descriptive statistics were calculated to describe the outbreak population and characterize the degree of environmental contamination. The Dice coefficient of similarity was calculatedl to assess relatedness among S. Newport isolates.

Results

Patient Population

In February 2004, in response to a perceived increase in salmonellosis, a retrospective and prospective epidemiological investigation was initiated. Historical data indicated that the proportion of Salmonella positive inpatients increased from 0% in 1998 to 3.8% between January 1 and May 10, 2004. Inpatients identified as high risk (presenting complaints of colic, diarrhea, or fever) had an even greater increase to 19.8% in 2004. The increased incidence could not be accounted for by changes in sampling frequency or alteration in the composition of the patient population over time. Although there was a 6% increase in inpatient admissions in 2003 (the year in which the index case was admitted) compared with 1998, the number of animals considered at high risk for shedding Salmonella was unchanged over the same time period. The sampling frequency (determined by the patient algorithm in place from 1986 until hospital closure in 2004 which consisted of passive patient surveillance) increased from 17 samples per 100 patient admissions in 1998 to 30 per 100 admissions in 2004. However, this difference does not account for the >16-fold increase in number of animals identified as positive.

On May 21, 2004, the index case infected with the outbreak strain of S. Newport MDR-Amp C was identified by epidemiological investigation and genotypic analysis. The animal was a 3-year-old female thoroughbred racehorse admitted to the hospital on July 1, 2003 with the primary complaint of colic. Therapeutic intervention involved surgical correction of a large colon volvulus and postoperative management of enterocolitis. The 1st sample from the index case to be identified as positive for S. Newport MDR-AmpC was collected on July 10, 2003.

All Salmonella-positive inpatients from July 1, 2003 until hospital closure were identified (N = 76). Results were obtained by passive surveillance (patient algorithm) of fecal (71/76, 93.4%), postmortem GI content (3/76, 3.9%), or blood (2/76, 2.6%) samples. Epidemic curves for positive inpatients are shown in Figure 1a and b; a timeline of major events during the outbreak is illustrated in Figure 1b. The distribution of positive patients by admission location is shown in Figure 2. During a 35-week period (July 2003–February 2004), patients positive for serotypes Newport (n = 28) and Typhimurium (n = 13) were identified. Over 10 weeks from March 1 until hospital closure, isolates identified as S. Newport MDR-AmpC were recovered from 33 patients; only 2 patients were positive for S. Typhimurium. This discovery, concomitant with PFGE data indicating genetic homogeneity among Newport isolates, provided definitive evidence of an outbreak.

Figure 1.

 (a) and (b) Epidemic curves for Salmonella positive inpatients identified between July 1, 2003 and May 10, 2004 at the George D. Widener Hospital for Large Animals. (a) shows the number of S. Newport MDR-AmpC (▪) and S. Typhimurium (▵) positive patients per month, plotted by the collection date of the 1st sample to be identified as positive, against the total number of inpatient admissions (•) for each month. In (b) the same positive patient data are shown plotted for each week of what was ultimately identified as the 45-week outbreak period; S. Newport MDR-AmpC (closed bars) and S. Typhimurium (open bars). Additionally, (b) depicts a timeline of main events of the outbreak:
1. July 1, 2003: Index case admitted.
2. July 10, 2003: Index case first positive.
3. Concerns over increased # of case fatalities develop over weeks.
4. Attempts to collect cases retrospectively by clinicians begins; difficulty obtaining information from medical records and microbiology.
5. March 10–28, 2004: Increased number of cases in ICU/NICU facility and orthopedic barn. Repeated cleaning and decontamination failed to prevent cases developing.
6. March 31, 2004: ICU/NICU facility and orthopedic barn closed for extensive decontamination and remediation.
7. April 6, 2004: Presentation of preliminary epidemiology and PFGE data at Morbidity / Mortality rounds.
8. April 9, 2004: ICU/NICU facility reopened.
9. April 13, 2004: Hospital closed to elective patient admissions for two weeks.
10. April 16, 2004: Phone consultation with biosecurity expert.
11. April 27, 2004: Samples collected and processed by MOD SURV; orthopedic barn reopened.
12. May 5–10, 2004: On-site visit from biosecurity experts.
13. May 10, 2004: Complete hospital closure.

Figure 2.

 Schematic of the George D. Widener Hospital for Large Animals, with distribution of Salmonella positive cases by admission location. Twenty-two cases were admitted directly to Isolation. Of the 54 remaining cases admitted elsewhere, 35 ultimately were moved to Isolation, although only 1 was sent directly; 34 had barrier precautions put in place stall-side. The remaining 19 positive patients were housed only in the main part of the hospital. Thirty-nine out of 76 positive patients occupied >1 stall. The maximum number of stalls occupied by a single positive patient was 5.

Sixty-one animals were infected with S. Newport MDR-AmpC. One patient had a dual infection with both S. Newport and S. Typhimurium; this horse was included as part of the outbreak population. Overall attack rates (cases/[cases + noncases] during outbreak period) were 2.1 and 1.0% in equine and nonequine patients, respectively. The median number of days from admission to first positive was 5 (IQR, 3–8 days; range, 0–43 days). The median number of samples submitted before the 1st positive sample identified was 0 (range, 0–4 samples). Considering that surveillance was passive and triggered by clinical signs only, this finding is not unexpected. Fifty-four horses were infected with the outbreak strain, in addition to 6 cattle and 1 lamb. The most prevalent clinical signs were inappetence, lethargy, fever, leukopenia, diarrhea, abdominal discomfort, and GI reflux. The case fatality rate for all species infected with S. Newport MDR-AmpC was 36.1% (22/61).

Among the affected equine population, a variety of presenting complaints were observed. Presenting complaint categories and associated attack rates are reported in Table 1. Attack rates were highest in horses presenting with enterocolitis and in foals and mares hospitalized for either suspected sepsis or PAS or enrolled in the high-risk pregnancy program. Horses were hospitalized for a median of 16 days (IQR, 9–22 days; range, 2–59 days). In severely affected positive patients, therapeutic interventions included aggressive antimicrobial therapy, IV fluid therapy, colloid support, administration of hyperimmunized and fresh frozen plasma, parenteral nutrition, and use of GI protectants. During the final phase of the outbreak (March–May 2004), defrayed bills for therapy and hospitalization, beyond that required for treatment of the presenting complaint, cost the hospital US$160,492. Despite aggressive therapy and minimal financial restrictions, many patients succumbed to the infection, resulting in a case fatality rate in horses of 31.5% (17/54).

Table 1.   Presenting complaint categories in all 54 equine patients subsequently identified as positive for Salmonella enterica serovar Newport MDR-AmpC.
Presenting Complaint
Categories (n = 54
Equine Patients)
Number of Horses
Identified as
S. Newport Positive
Attack Ratea
over Outbreak
Period (%)
  • Patients are stratified into presenting complaint categories according to presumed risk for developing nosocomial salmonellosis. Arthroscopic, endoscopic, or orthopedic surgical patients, as well as a horse presenting with presumed neurologic disease, were categorized as “other.” The suspected sepsis/PAS category comprised foals presenting to the NICU with either suspected sepsis or PAS.

  • ICU, intensive care unit; NICU, neonatal intensive care unit; PAS, perinatal asphyxia syndrome; MDR, multidrug resistant.

  • a

    Attack rates for “Companion” and “Other” patient categories are combined, as are all mare/foals admitted to the High-Risk Pregnancy program or ICU/NICU with presumed sepsis or PAS.

Gastrointestinal nonenterocolitis145.3
Enterocolitis816.3
Companion90.9
Other8 
Suspected sepsis/PAS5 
High-risk pregnancy-foal78.5
High-risk pregnancy-mare3 

Environmental Contamination

Routine environmental surveillance (ORIG SURV) results were available from 722 samples collected between January 1998 and March 3, 2004; 24 (3.3%) were positive. There was serogroup information on 18 of the 24 isolates; 10 were serogroup C2, 7 serogroup B, and 1 was serogroup D. Ten (41.7%) of the positive samples were recovered after admission of the index case on July 1, 2003 (7 were serogroup C2, 3 were serogroup B).

Environmental surveillance by the ORIG SURV technique increased on March 10, 2004 in areas of the hospital in which salmonellosis had occurred (171 samples collected). From March 10, 2004 until closure, 15 (8.8%) samples collected from the ICU/NICU area and orthopedic barn were positive. As a result, each area was depopulated and subjected to extensive cleaning, decontamination, and facility remediation. Environmental samples collected after this time (ORIG SURV) were negative. Nevertheless, an additional 5 cases of salmonellosis occurred in these areas. When a more sensitive, modified ISO culture method (a component of MOD SURV) was implemented at the end of April, 8/25 (32%) environmental samples were positive. At this time, the ICU/NICU and orthopedic barn were closed to new admissions.

Salmonellosis also increased among patients housed in other areas of the hospital, but environmental culture results in these barns and the clinical procedures area remained negative (ORIG SURV). However, MOD SURV results from the 120 samples collected (April 27, 2004) in these areas indicated widespread environmental contamination; 36/120 (30%) were positive for Salmonella serogroup C2. Human traffic and animal areas were affected; all isolates were later confirmed as S. Newport MDR-AmpC. Confirmation of extensive and repeated contamination, combined with findings from a site visit by biosecurity experts (Drs Paul Morley and David Van Metre, May 5–10, 2004) resulted in full hospital closure on May 10, 2004.

After closure, 222 samples were collected from across the extended campus (650 acres outside of the main hospital); only 3 were positive for Salmonella, and none matched the outbreak strain. Based on these results, restricted (Widener Hospital) and nonrestricted (extended campus) zones were identified and traffic control measures were implemented. Only personnel involved in cleaning and decontamination were permitted in the hospital area. Physical barriers were placed directing human traffic through single entrances with foot baths and mats. Certain functions (eg, research, farm, diagnostic laboratories, and reproductive services) confined to noncontaminated areas of the campus operated normally. Temporary fences were placed to separate contaminated and noncontaminated areas of the campus. The school's ambulatory and outpatient services were relocated to nonrestricted zones and continued to serve the community and provide professional student rotations while the hospital was closed.

Research and Teaching Animals

Samples were collected from the 345 research and teaching animals housed on the NBC campus. One hundred and sixty-eight individual or pooled samples were examined. With the exception of a group E Salmonella isolated from a single swine sample, no Salmonella were isolated from any other animals.

Strain Description

All patient isolates were identified by serotyping as S. Newport and shown to be resistant to ampicillin, chloramphenicol, tetracycline, cephalothin, ceftiofur, amoxicillin-clavulanic acid, gentamicin, and trimethoprim-sulfamethoxazole; susceptible to amikacin and imipenem; and either susceptible or intermediate to enrofloxacin. Isolates from 40/61 patients were available for molecular characterization by PFGE and 10 unique, but highly related (Dice coefficient of similarity >94%), XbaI profiles were observed from the 40 isolates. Two PFGE profiles, NP102 (17 isolates) and NP109 (9 isolates), predominated (Figure 3). The outbreak-related strains showed ≤88% similarity to the predominant S. Newport MDR-AmpC strain of Pennsylvania origin in a database of 990 isolates obtained from the PADLS in the 18 months that preceded the outbreak. A full molecular characterization of the outbreak strain has been reported previously and the presence of blaCMY-2, blaTEM-1b, and blaSHV-12 genes was identified in SRC0307-213 (index case).28blaSHV-12 encodes an extended spectrum beta-lactamase (ESBL). Salmonella strains that express ESBLs are rare, and to our knowledge, this was the 1st example of an ESBL gene in S. Newport MDR-AmpC.28

Figure 3.

 Dendrogram demonstrating the 9 different distinct pulsed-field profiles that were identified during the outbreak.

Facility Remediation and Decontamination

An interim Director of Biosecurity was appointed after hospital closure to manage facility remediation and decontamination. All animal housing and clinical spaces were subjected to a 3 or 4 stage cleaning and disinfection protocol (Table 2). Cleaned areas were closed to traffic except for the collection of environmental samples (MOD SURV) to verify negative status. More than 220,000 square feet of the facility were cleaned in this manner by faculty members, house officers, and hospital staff.

Table 2.   Four-stage cleaning and disinfection protocol used in all contaminated areas of the hospital.
StageAgentMethod of Application and Contact TimePreparation for Next Stage
  1. Stage 4 was not used in the ICU/NICU, which required gas phase decontamination.

  2. ICU, intensive care unit; NICU, neonatal intensive care unit.

1Anionic detergentmHand-held brush 15 minutesRinse, dry
22% bleachnMop/sponge applicator 20 minutesRinse, dry
3Quaternary ammonium disinfectantoMop/pump sprayer 5 h/overnightRinse, dry
44% Peroxygen-based disinfectantpApplication by fogging30,31No rinse, dry

Extensive remediation was deemed necessary because of the nature of the contamination and the inability to disinfect damaged surfaces. The 4 main cement-block barns were sandblasted and resurfaced, cleaned, disinfected, and repainted. All dirt stall flooring (17 stalls in the orthopedic barn and 1 down-cow stall) was removed and replaced with concrete. Stall mats in the remaining 80 box stalls also were removed. A polyurethane-based monolithic flooring systemq was installed in all stalls and animal handling areas. Remediation of human traffic areas within clinical spaces included removing damaged equipment and replacing noncleanable surfaces.

The ICU/NICU was a widely contaminated facility with a sophisticated infrastructure, including an extensive air handling system. After a 3-stage liquid-phase cleaning, the ICU/NICU was decontaminated by professional contractorsr,s with chlorine dioxide as a gas-phase decontaminant.32 Decontamination was confirmed by obtaining negative growth from 3 different biological indicators placed at 40 sites throughout the building. The indicators were commercially prepared strips inoculated with approximately 2 × 106 spores of Bacillus atrophaeust or Geobacillus stearothermophilist and strips prepared at the SRC inoculated with approximately 109 colony forming units of S. Newport MDR-AmpC.

Financial Impact of Outbreak, Closure, and Remediation

Average hospital income by month for the 5 years before the outbreak was calculated. Over the 11 months in which revenue was decreased by >2 standard deviations from the 5-year monthly average (May 2004–March 2005), estimated lost revenue caused by closure, staged reopening, and decreased caseload totaled US$3.25 million. Monthly revenue did not return to previously generated income until April 2005. Cost of facility remediation and decontamination totaled US$704,574. This included replacement of drains and flooring (35.2%); sandblasting, painting, and other work in barns (34.9%); replacement of equipment and supplies (11.3%); ICU/NICU gas phase decontamination and rebalancing of the air handling system (8.1%); sample collection and culturing (7.3%); and costs for personal protective equipment, cleaning supplies, and materials disposal (3.2%). Revenue lost because of hospital coverage of patient bills was US$160,492, bringing the total financial impact of the outbreak and subsequent closure to approximately US$4.12 million.

Implementation of ICP

After successful decontamination, staged hospital reopening began in August 2004. Reopening of the ICU/NICU in January 2005 signaled the return to full operation. A full commitment to biosecurity was made at the highest level of the school. A Director of Biosecurity was charged with developing a long-term biosecurity plan with the assistance of a Biosecurity Advisory Committee that included representatives from all clinical services, critical support staff, and diagnostic services. A dramatically modified ICP incorporating sampling of all inpatients throughout hospitalization and extensive environmental surveillance, risk stratification, and traffic control of patients and people, adherence to excellent hygiene practices, and stakeholder education was implemented. This represented a major paradigm shift in pre-existing attitudes regarding infection control in the entire veterinary school.

Discussion

The outbreak at the Widener Hospital caused by S. Newport MDR-AmpC lasted 10 months, involved many hospital patients, and had high case fatality. This report demonstrates that a nosocomial outbreak can occur despite an existing ICP. Failures in the previous ICP identified during this investigation include lack of a dedicated individual responsible for biosecurity, reliance on passive and not comprehensive patient surveillance (active and algorithm-driven components), poor enforcement of and compliance with existing protocols, no centralized database for analysis of patient or environmental results, no real-time data evaluation, and no modifications to the program in response to changes in microbial threats. Patients were not housed based on risk for nosocomial infection (ie, risk stratification), and there were few explicit directives for implementation of barrier precautions in areas other than isolation. There was no definition of what constituted nosocomial disease in the previous ICP.

Because of the passive approach to patient surveillance in place at the time of the outbreak, the number of animals identified as infected likely represents only a portion of patients that may have become colonized during the epidemic. Before the outbreak, patient surveillance was based on a loosely followed patient algorithm that targeted animals with diarrhea only. Routine surveillance of patients, either healthy or at increased risk for colonization or infection with Salmonella, was not performed. Additionally, absence of information regarding incidence of Salmonella shedding in various subgroups of the patient population led to inappropriate mixing of patients from different risk categories. The most obvious example was housing GI patients in the ICU/NICU, most likely contributing to the contamination of a difficult-to-clean facility. This may have been a factor in the high attack rate observed in the mare and foal population housed in the ICU/NICU during the 2004 foaling season. In addition, GI cases were routinely triaged to a barn housing outpatients and patients hospitalized for elective procedures. This approach probably exacerbated widespread contamination.

The methods for environmental surveillance in place before recognition of the outbreak (ORIG SURV) were flawed, from both a collection and a culture technique standpoint. ORIG SURV techniques used for the collection of environmental samples led to the false belief that the hospital was not contaminated. Despite the perceived increase in clinical cases, nosocomial Salmonella was considered less likely than community-acquired infection because there was no identified increase in environmental contamination. Advantages of the MOD SURV sample collection technique include wider recovery of debris and bacteria because of the electrostatic charge on the wipes, as well as the ability to sample larger areas collectively. Modifications to the culture protocol aimed at bacterial resuscitation and differential selection for Salmonella (modified ISO technique) facilitated recovery of Salmonella organisms that may have been present in the environment, but were either debilitated or overgrown by competing bacteria. Implementation of the MOD SURV technique after biosecurity consultation was essential in determining the degree of hospital contamination. The appearance of cases of S. Typhimurium from November 2003 through February 2004 increased concerns over salmonellosis, but the involvement of 2 serovars (Newport and Typhimurium) was regarded as further evidence against a nosocomial outbreak.

The age and poor cleanability of much of the hospital made rapid disinfection and decontamination more challenging, and made remediation costly. The need for chloride dioxide gas decontamination in a facility with a sophisticated air handling system increased the cost of outbreak remediation. Infection control and ease of disinfection are essential considerations in the design or renovation of a VTH.

Unique features of the causative organism, S. enterica subspecies enterica serovar Newport, include marked environmental persistence and MDR. Prolonged environmental survival of S. Newport, including MDR-AmpC strains, has been described in several studies33,34 and has been observed as a feature of a number of outbreaks (Dr Robert Munson, University of Pennsylvania and PADLS Field Investigation, personal communication). Repeated, aggressive decontamination of areas proved unsuccessful, and salmonellosis recurred in affected areas. Increased antimicrobial resistance was mediated by expression of an ESBL gene combined with the presence of a plasmid-borne ampC gene (blacmy2) that conferred resistance to extended-spectrum cephalosporins.28 The emergence of bacteria carrying multiple β-lactamase genes has important implications for veterinary and public health; the existence of certain genetic combinations could effectively eliminate all β-lactam therapeutic options.

A commitment to biosecurity at the highest level within the school and the subsequent development of an effective ICP were important components to successful hospital reopening. Dedicated faculty and staff positions with sole responsibility of infection control were created (Director of Biosecurity, Biosecurity Assistant). The sensory input to the revised ICP consisted of rigorous fecal surveillance of all inpatients, with the aim of establishing incidence of Salmonella shedding in various segments of the hospital's patient population. Implementation of a more rigorous patient algorithm35 triggers a mandatory series of 3 fecal cultures, implementation of barrier precautions, and strict patient relocation to isolation if specific criteria are met. Clinical status of patients in high risk areas is monitored by the biosecurity staff in liaison with attending clinicians. Upon reopening, environmental surveillance was performed weekly in all areas of the hospital (34 samples/wk) by the MOD SURV technique, with additional samples as needed. Failure of faculty, staff, and students to comply with the ICP results in repercussions from administrative leaders.

The nosocomial outbreak reported here demonstrates the importance of an effective ICP in a VTH, and emphasizes the impact of a severe, protracted outbreak characterized by extensive facility contamination and high case fatality. The financial impact of a major outbreak causing hospital closure and extensive remediation also is emphasized. The financial consequences of revenue loss, decontamination, and facility remediation provide valuable information to those evaluating the cost : benefit ratio of establishing a new ICP or revising an existing program. The effect of such an outbreak on public sentiment toward an institution cannot be easily quantified, but also is important.

Footnotes

aWhirl-pak sterile sample collection bag, Nasco Inc, Modesto, CA

bSelenite broth, Difco, Becton Dickinson and Company, Sparks, MD

cElectrostatic Wipe, Swiffer, Procter and Gamble, Cincinnati, OH

dSweeper, Swiffer, Procter and Gamble

eXLD agar plates, Fisher Scientific, Pittsburgh, PA

fMacConkey II agar plates, Fisher Scientific

gSerogrouping antisera, Difco, Becton Dickinson and Company

hSensititre CMV2ECOF Companion/Equine MIC Veterinary Specific plate, and Food Animal Bovine/Porcine panel BOPO6F, Trek Diagnostics, Cleveland, OH

iSerotyping antisera, Statens Serum Institute, Fisher Scientific

jBBL Septi-Chek, Becton Dickinson and Company

kKodak EDAS120, Life Technologies, Gaithersburg, MD

lBioNumerics software, Applied Maths, Kortrijk, Belgium

mA-125, Ecolab, St Paul, MN or Tide, Procter and Gamble

nClorox, Clorox Co, Oakland, CA

oA-456-N, Ecolab

pVirkon-S, Antec International, Sudbury, UK applied using a Solo 450 motorized mist blower; Solo, Newport News, VA

qAbacus Sports Installations Ltd, Lancaster, PA

rMicro-Clean Inc, Bethlehem, PA

sChlorDiSys Solutions Inc, Lebanon, NJ

tSpordex, Steris Corporation, Mentor, OH

Acknowledgments

This work was supported in part by funding from the Pennsylvania Department of Agriculture.

Ancillary