Epidemiology, Surveillance and Infection Control Aspects of Salmonella Infections in Hospitalized Horses

MISSING IMAGE

Material Information

Title:
Epidemiology, Surveillance and Infection Control Aspects of Salmonella Infections in Hospitalized Horses
Physical Description:
1 online resource (197 p.)
Language:
english
Creator:
Ekiri, Abel B
Publisher:
University of Florida
Place of Publication:
Gainesville, Fla.
Publication Date:

Thesis/Dissertation Information

Degree:
Doctorate ( Ph.D.)
Degree Grantor:
University of Florida
Degree Disciplines:
Veterinary Medical Sciences, Veterinary Medicine
Committee Chair:
Hernandez, Jorge A
Committee Members:
Mackay, Robert J
Morton, Allison
Long, Maureen T
Krueger, Traci M.

Subjects

Subjects / Keywords:
colic -- diarrhea -- epidemiology -- horse -- hospital -- pcr -- salmonella -- surveillance
Veterinary Medicine -- Dissertations, Academic -- UF
Genre:
Veterinary Medical Sciences thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract:
This research work addressed four objectives. The first part was a critical review of published studies that have contributed to the literature of epidemiology and infection control of nosocomial Salmonella infections in hospitalized horses. This review identified several knowledge gaps related to epidemiology, diagnostics, and surveillance and infection control of Salmonella infections in hospitalized horses. The review was followed by three investigations that addressed three of the identified knowledge gaps. The second part was an investigation of the relationships between clinical parameters (clinical signs and procedures, hematological and plasma chemistry) and Salmonella shedding in horses with signs of GI disease and with or without diarrhea. This study was designed as a case-control study. Results revealed that high plasma triglycerides at admission, abdominal surgery, and season (summer) were associated with Salmonella shedding in horses without diarrhea. These parameters can be considered for evaluation in hospital biosecurity programs to identify equine GI inpatients at high risk of Salmonella shedding. The third part was an investigation that assessed the diagnostic performance of a real-time polymerase chain reaction (PCR) assay for detection of Salmonella spp in feces of hospitalized horses, compared to bacteriological culture. Fecal samples from horses classified as true-positives and true-negatives were used to assess the relative sensitivity and specificity of the PCR assay, respectively. Two primer sets were evaluated and results revealed that this PCR protocol was reliable, with a relatively high specificity (86-100%) and sensitivity (100%) when compared to bacteriological culture. This real-time PCR protocol can be used as a surveillance tool to detect Salmonella spp in fecal specimen of hospitalized horses. The fourth part was a survey that assessed the awareness and relevance of a hospital surveillance and infection control program among referral veterinarians and clients who refer or send horses to a referral hospital for veterinary care. A total of 92 RDVMs and 594 clients participated in the survey. Study results revealed that most clients (61%) were not aware that the referral hospital operates an infection control program, and both RDVMs and clients considered it very important that a referral hospital operates a surveillance and infection control program.
General Note:
In the series University of Florida Digital Collections.
General Note:
Includes vita.
Bibliography:
Includes bibliographical references.
Source of Description:
Description based on online resource; title from PDF title page.
Source of Description:
This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility:
by Abel B Ekiri.
Thesis:
Thesis (Ph.D.)--University of Florida, 2012.
Local:
Adviser: Hernandez, Jorge A.
Electronic Access:
RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2013-05-31

Record Information

Source Institution:
UFRGP
Rights Management:
Applicable rights reserved.
Classification:
lcc - LD1780 2012
System ID:
UFE0044081:00001


This item is only available as the following downloads:


Full Text

PAGE 1

1 EPIDEMIOLOGY, SURVEILLANCE AND INFECTION CONTROL ASPECTS OF SALMONELLA INFECTIONS IN HOSPITALIZED HORSES By ABEL B. EKIRI A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMEN T OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2012

PAGE 2

2 2012 Abel B. Ekiri

PAGE 3

3 To my parents, Ms. Josephine Birabwa and the late Dr. Richard Bulamu, that unreservedly contributed to the foundation of my educatio n.

PAGE 4

4 ACKNOWLEDGMENTS I thank my graduate committee members, Drs. Jorge Hernandez, Traci Krueger, Robert Mackay, Alison Morton and Maureen Long for their remarkable support. I am particularly appreciative of Dr. Long for her support with the molecular aspects of this research work and my major professor, Dr Hernandez for his tremendous support and guidance throughout my graduate studies. I extend my gratitude to the college of veterinary medicine, department of large animal clinical sciences, and the l arge animal hospital for funding my graduate studies. I am indebted to several friends and colleagues that contributed in various ways to my academic success. Thank you for the advice and constructive criticism, moral and financial support. Lastly, I wou ld like to thank my family for the continued love, support and encouragement through this academic journey.

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 8 LIST OF FIGURES ................................ ................................ ................................ ........ 10 LIST OF ABBREVIATIONS ................................ ................................ ........................... 11 ABSTRACT ................................ ................................ ................................ ................... 12 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .... 14 Research Objectives and Approach ................................ ................................ ....... 17 Objective 1: Review of Epidemiology and Infection Control Aspects of Nosocomial Salmonella Infections in Hospitalized Horses ............................ 17 Objective 2: Exposure Factors Associated with Salmonella Shedd ing among Hospitalized Horses with or without Diarrhea ................................ .... 17 Objective 3: Diagnostic Performance of Real Time PCR for Diagnosis of Salmonella Shedding in Hospitalized Horses ................................ ................ 18 Objective 4: Awareness and Relevance of Hospital Surveillance and Infection Control Services among Referral Veterinarians and Clients ........... 19 2 LITERAT URE REVIEW ................................ ................................ .......................... 20 Etiology ................................ ................................ ................................ ................... 20 Pathogenesis ................................ ................................ ................................ .......... 20 Salmonella Virulence M echanisms ................................ ................................ ......... 24 Salmonella Virulence Genes Required for Invasion of Intestinal Epithelial Cells ................................ ................................ ................................ .............. 26 Survival of Salmonella Following Intestinal Invasion ................................ ........ 29 Multi Drug Resistant Salmonella Strains in Horses ................................ .......... 30 Host Immunity against Salmonella Infections ................................ ......................... 32 Epidemiology and Infection Control Aspects of Nosocomial Salmonella Infections in Hospitalized Horses ................................ ................................ ......... 35 Surveillance and Infection Control ................................ ................................ .... 36 Case definition for nosocomial Salmonella infections ................................ 36 Environmental contamination as a source of nosocomial Salmone lla infections in hospitalized horses ................................ ............................. 40 Sampling ................................ ................................ ................................ .... 42 Laboratory diagnosis ................................ ................................ .................. 43 Risk management ................................ ................................ ...................... 50 Guidelines for surveillance and infection control ................................ ........ 52

PAGE 6

6 Epidemiologic Research ................................ ................................ ................... 59 Discussion ................................ ................................ ................................ ........ 62 3 EXPOSURE FACTORS ASSOCIATED WITH SALMONELLA SHEDDING AMONG HOSPITALIZED HORSES WITH OR WITHOUT DIARRHEA .................. 71 Materials and Methods ................................ ................................ ............................ 73 Study Population ................................ ................................ .............................. 73 Hospital Surveillance and Infection Contr ol Procedures ................................ ... 73 Study Design ................................ ................................ ................................ .... 74 Selection of cases ................................ ................................ ...................... 75 Selection of controls ................................ ................................ ................... 76 Microbiologic Procedures for Detection of Salmonella Spp .............................. 76 Data Collection ................................ ................................ ................................ 77 Data Analysis ................................ ................................ ................................ ... 78 Results ................................ ................................ ................................ .................... 80 Model 1: Case Horses that Tested Positive for Salmonella Admission and had Diarrhea ................................ ................................ ......... 82 Model 2: Case Horses that Tested Positive for Salmonella Admission and did not have Diarrhea ................................ ........................... 83 Model 3: Case Horses that Tested Positive for Salmonella Admission and did not have Diarrhea, and Tested Negative on at least 2 ................................ ...... 83 Discussion ................................ ................................ ................................ .............. 85 4 DIAGNOSTIC PERFORMANCE OF REAL TIME PCR FOR DIAGNOSIS OF SALMONELLA SHEDDING IN HOSPITALIZED HORSES ................................ ... 116 Materials and Methods ................................ ................................ .......................... 119 Study Population ................................ ................................ ............................ 119 Study Design ................................ ................................ ................................ .. 120 Fecal Sample Collection ................................ ................................ ................. 121 Microbiological Procedures ................................ ................................ ............ 122 Real time PCR Assay Procedures ................................ ................................ 123 Bioinformatics analysis ................................ ................................ ............ 123 Optimization of primer and probe concentration ................................ ...... 124 Analytical specificity and sensitivity of the PCR assay ............................. 125 Comparison of the analytical sensitivity of DNA extraction kits ................ 126 Evaluation of PCR assay detection limit ................................ .................. 128 Detection of Salmonella DNA in study samples ................................ ....... 129 Data Collection ................................ ................................ ............................... 130 Data Analysis ................................ ................................ ................................ 130 Results ................................ ................................ ................................ .................. 131 Bioinformatics Analysis ................................ ................................ ................... 131 Optimization of Primer and Probe Concentration ................................ ........... 133 Analytical Specificity and Sensitivity ................................ ............................... 134 Comparison of the Analytical Sensitivity of DNA Extraction Kits .................... 134

PAGE 7

7 PCR Assay Detection Limit ................................ ................................ ............ 135 Detection of Salmonella DNA in Study Samples ................................ ............ 135 Relative Specificity ................................ ................................ ................... 136 Relative Sensitivity ................................ ................................ ................... 137 Discussion ................................ ................................ ................................ ............ 138 5 AWARENESS AND RELEVANCE OF HOSPITAL SURVEILLANCE AND INFECTION CONTROL SERVICES AMONG REFERRAL VETERINARIANS AND CLIENTS ................................ ................................ ................................ ...... 158 Materials and Methods ................................ ................................ .......................... 159 Study Population ................................ ................................ ............................ 159 Study Design ................................ ................................ ................................ .. 160 Data Collection ................................ ................................ ............................... 161 Data Analysis ................................ ................................ ................................ 163 Results ................................ ................................ ................................ .................. 163 General Information for Referral DVMs ................................ .......................... 164 General Information for Clients ................................ ................................ ....... 164 Awareness ................................ ................................ ................................ ...... 165 Relevance ................................ ................................ ................................ ...... 165 Discussion ................................ ................................ ................................ ............ 166 6 SUMMARY AND RESEARCH CONCLUSIONS ................................ ................... 173 APPENDIX A SURVEY FOR REFERRAL DVMS ................................ ................................ ....... 177 B SURVEY FOR CLIENTS ................................ ................................ ...................... 179 LIST O F REFERENCES ................................ ................................ ............................. 181 BIOGRAPHICAL SKETCH ................................ ................................ .......................... 197

PAGE 8

8 LIST OF TABLES Table page 2 2 Large animal hospitals that have closed because of an outbreak of nosocomial Salmonella infections and surveillance parameters of interest: ....... 68 3 1 Salmonella serotypes isolated from cases with or without d iarrhea .................... 93 3 2 Univariable unconditional logistic regression analysis of investigated exposure factors at admission associated with Salmonella after admission in horses with diarrhea ................................ .............................. 94 3 3 Univariable unconditional logistic regression analysis of investigated exposure factors at admission associated with Salmonella after admission in horses with out diarrhea ................................ ....................... 100 3 4 Univariable unconditional logistic regression analysis of investigated exposure factors (during hospitalization on days 2 and 3) associated with Salmonella ys after admission in horses without diarrhea ....... 106 3 5 Multivariable analysis of investigated exposure factors at admission associated with Salmonella s without diarrhea ................................ ................................ ................................ 114 3 6 Multivariable analysis of investigated exposure factors (during hospitalization) associated with Salmonella admission in horses without diarr hea ................................ ............................... 115 4 1 Primer optimization matrix used to determine the minimum primer concentration that yielded the minimum C T ...................... 146 4 2 A summary of nucleotide blast results of published individual primers and probes ................................ ................................ ................................ .............. 147 4 3 Nucleotide sequences of the two selected sets of primers and probes and universal bacter ia primer and probe ................................ ................................ 148 4 4 A summary of nucleotide blast results for the invA and histidine transport operon genes ................................ ................................ ................................ .... 149 4 5 Mean C T values for the DNeasy tissue kit, UltraClean fecal kit, and QIAamp DNA Stool Mini Kit for 3 trials performed in triplicates using Primer set 1. ....... 150 4 6 Mean C T values for the DNeasy tissue kit, UltraClean fecal kit, and QIAamp DNA Stool Mini Kit for 3 trials performed in triplicates using Primer set 2. ....... 150

PAGE 9

9 4 7 Relative specificity of the real time PCR assay determined using fec al samples from non GI horses that tested culture negative on five consecutive samples ................................ ................................ ................................ ............ 15 1 4 8 Relative specificity of the real time PCR assay determined using fecal samples from GI horses tha t tested culture negative on atleast three samples 151 4 9 Bacteriological culture and real time PCR results for fecal samples collected from GI horses that tested Salmonella positive and were classified as true positives (n=20) ................................ ................................ ................................ 151 5 1 Survey responses from RDVMs: general information ................................ ....... 169 5 2 Survey responses from clie nts: general information ................................ ......... 170 5 3 Survey responses from RDVMs and clients: awareness ................................ .. 171 5 4 Survey responses from RDVMs and clie nts: relevance ................................ .... 172

PAGE 10

10 LIST OF FIGURES Figure page 2 1 A diagram showing two potential sources of nosocomial infection: an inpatient index case or envi ronmental contamination. ................................ ........ 69 2 2 Surveillance and infection control communication channels at the University of Florida, Large Animal Hospital. ................................ ................................ ....... 70 4 1 Sampling approach used to estimate the relative sensitivity and specificity of the real time PCR assay. ................................ ................................ ................. 152 4 2 Dissociation curve analysis for primer set 1 using SYBR Gr een: showing one melting peak for the primer concentrations that were evaluated, indicating absence of non specific primer binding. ................................ ........................... 153 4 3 Dissociation curve analysis for primer set 2 using S YBR Green: showing one melting peak for the primer concentrations that were evaluated, indicating absence of non specific primer binding. ................................ ........................... 154 4 4 Dissociation curve analysis for the universal bacteria primer set using SYBR Green: showing one melting peak for the primer concentrations that were evaluated, indicating absence of non specific primer binding. .......................... 155 4 5 Standard curve for primer set 1 ................................ ................................ ........ 156 4 6 Standard curve for primer set 2 ................................ ................................ ........ 157

PAGE 11

11 LIST OF ABBREVIATION S CFU Colony forming unit CT Number of cycles required for the fluorescent signal to cross the threshold C Degree Celsius Delta normalized reporter value of an experimental reaction minus the normalized reporter value of the baseline signal generated by the instrument F Fahrenheit fL Femtoliters GI Gastrointestinal g/dL Grams per deciliter K/uL Thousands per cubic milliliter mEq/L Milliequivalents per litre solvent mg/dL Milligrams per dec iliter mmol/L Millimoles per liter M/ u L Millions per cubic milliliter mL Milliter pg Picograms PCR Polymerase chain reaction Spp Species U /L Units per liter

PAGE 12

12 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy EPIDEMIOLOGY, SURVEILLANCE AND INFECTION CONTROL ASPECTS OF SALMONELLA INFECTIONS IN HOSPITALIZED HORSES By Abel B. Ekiri May 2012 Chair: Jorge Hernandez Major: Veterinary Medical Sciences This research work addressed four objectives The f irst part was a critical review of published studies that h ave contributed to the literature of epidemiology and infection control of nosocomial Salmonella infections in hospitalized horses. This review identified several knowledge gaps related to epidemiology, diagnostics, and surveillance and infection control o f Salmonella infections in hospitalized horses. Th e review was followed by three investigat ions that address ed three of the identified knowledge gaps The second pa rt was an investigation of the relationships between clinical parameters ( clinical signs an d procedures hematological and plasma chemistry ) and Salmonella shedding in horses with signs of GI disease and with or without diarrhea. This study was designed as a case control study. Results revealed that high plasma triglyceride s at admission abdomi nal surgery and season (summer) were associated with Salmonella shedding in horses without diarrhea. These parameters can be considered for evaluation in hospital biosecurity programs to identify equine GI inpatients at high risk of Salmonella shedding. The third part was an investigation that assess ed the diagnostic performance of a real time polymerase chain reaction ( PCR ) assay for detection of Salmonella spp in

PAGE 13

13 feces of hospitalized horses, compared to bacteriological culture. Fecal samples from horse s classified as true positives and true negatives were used t o assess the relative sensitivity and specificity of the PCR assay respectively. Two primer sets were evaluated and results revealed that this PCR protocol was reliable, with a relatively high s pecificity (86 100%) and sensitivity (100%) when compared to bacteriological culture. This real time PCR protocol can be used as a surveillance tool to detect Salmonella spp in fecal specimen of hospitalized horses. The fourth part was a survey that ass ess ed the awareness and relevance of a hospital surveillance and infection control program among referral veterinarians and clients who refer or send horses to a referral hospital for veterinary care. A total of 92 RDVMs and 594 clients participated in the survey S tudy results revealed that most clients (61%) were not aware that the referral hospital operates an infection control program, and both RDVMs and clients considered it very important that a referral hospital operates a surveillance and infection control program.

PAGE 14

14 CHAPTER 1 INTRODUCTION Shedding of Salmonella spp in feces of hospitalized horses is an important problem for large animal hospitals because of the potential risk of nosocomial Salmonella infections. Consequences of outbreaks of nosoco mial Salmonella infections can be significant and may include high mortality in equine inpatients and substantial financial losses. 1 6 Because of the potential risk of outbreaks of nosocomial Salmonella infections, several large animal hospitals have estab lished surveillance and infection control programs to allow early detection of Salmonella shedding in hospitalized animals and timely implementation of appropriate preventative and control measures. T he frequency of horses shedding Salmonella during hospi talization can be higher among horses with signs of gastrointestinal (GI) tract disease without diarrhea compared to horses with diarrhea. Salmonella positive horses without diarrhea present a threat to veterinary hospitals because bacteriological culture procedures for detection of Salmonella shedding are often completed after the patient has been discharged, causing a delayed response in implementation of infection control measures. Several epidemiologic studies have been conducted to identify exposure fa ctors associated with Salmonella shedding in hospitalized horses. However in those studies, the analyses did not separate horses with diarrhea from those without diarrhea. As a result, the reported risk factors apply to a combined population of horses with and without diarrhea. Exposure factors associated with Salmonella shedding in horses with or without diarrhea have not been independently investigated. This information is important as it can further improve current hospital surveillance programs for ear ly detection of Salmonella shedding and allow rapid implementation of infection control measures.

PAGE 15

15 Several PCR protocols have been developed and evaluated for potential use in hospital surveillance programs for early detection of Salmonella spp in feces of hospitalized horses. The advantage of using PCR protocols to monitor Salmonella shedding in feces of hospitalized horses is that test results can be obtained more quickly, compared to bacteriological cu 5 days), thereby allowing an earlier implementation of preventative measures such as isolation and barrier nursing that minimize the risk of a potential outbreak of nosocomial Salmonella infection. 7,8 However, despite the availab ility of rapid and cost effective PCR tests for detection of Salmonella its acceptance as a surveillance tool for monitoring community acquired and nosocomial Salmonella infections in horses has been limited. The main reasons for not using PCR testing as a surveillance tool are that in previous studies 9 13 sampling methods used to establish the gold standard for true positives and true negatives were not objective T his led to potential misclassification bias and inconclusive results In particular, ther e was a high frequency of Salmonella PCR false positive results in horses without clinical signs of salmonellosis that test ed negative to Salmonella spp by bacteriological culture on multiple fecal samples These issues were perhaps due to the use of prime rs that were non specific or cross react ive with other bacteri a For PCR to be considered as a surveillance tool, these limit ations need to be addressed through use of correct sampling approaches, consistent standard definitions for true positives and true negatives, and use of a real time PCR protocol with primers and probes that have demonstrated no evidence of cross reactivity with other enteric and non enteric organisms.

PAGE 16

16 At a referral hospital referring veterinarians and clients play a vital role i n a hospital surveillance and infection control program Veterinarians p articipate in the program by referring clients to a tertiary care hospital because of its specialized expertise and need for specialized care. In turn, clients bring high risk patients su ch as those with signs of gastrointestinal (GI) disease that may be targeted in a hospital surveillance and infection control program At a referral hospital, information about a hospital survei llance and infection control program and its importance to RDV Ms and clients is promoted in various ways. At the UF LAH, educational efforts about the program include provision of an infection control brochure to all new clients to educate them about the infection control measures that are instituted at the hospital to optimize patient care. One element of an animal disease surveillance system is evaluation and feedback. To have an efficient program, it i s important that performance of the program is evaluated and feedback is collected from all individuals involved in the program including clinicians, hospital personnel, RDVMs and clients. Considering the role played by RDVMs and clients, it is important to know the level of awareness and perceptions about the relevance of a hospital surveillance and infection program among this group This information can be used to guide decision making on hospital policy issues related to surveillance and infection control and streamlin e costs and efficiency of services provided to clients.

PAGE 17

17 Research Objectives and Approach Objecti ve 1: Review o f Epidemiology a nd Infection Control Aspects o f Nosocomial Salmonella Infections i n Hospitalized Horses The purpose of this objective was to review published studies that have contributed to the literature of epidemiology and infection contro l aspects of nosocomial Salmonella infections in hospitalized horses. The review is presented as part of Chapter 2 and was structured in 3 parts: 1) surveillance and infection control; 2) epidemiological research; and 3) discussion. The first part consists of six elements case definition, environmental contamination, sampling, laboratory diagnosis, risk management, and guidelines for surveillance and infection control. The second part was limited to a review of published studies that examined exposure facto rs associated with nosocomial Salmonella infections in hospitalized horses. Finally, the third part was a discussion section relevant to infection control issues and areas of research that can further improve existing hospital surveillance and infection co ntrol programs. Objective 2: Exposure Factors Associated w ith Salmonella Shedding a mong Hospitalized Horses w ith o r w ithout Diarrhea The objective of this study was to investigate the relationships between clinical signs, hematological and plasma chemistry parameters, and clinical procedures (before admission, at admission, and during hospitalization) and Salmonella shedding in hospitalized horses with signs of GI disease, with or without diarrhea. A case control study approach was used to identify investig ated exposure factors associated with Salmonella shedding. The study population included adult horses with signs of GI disease and the sample of study animals included case horses with diarrhea, case horses without diarrhea, and control horses.

PAGE 18

18 Objective 3 : Diagnostic Performance o f Real Time PCR f or Diagnosis o f Salmonella Shedding i n Hospitalized Horses The objective of this study was to assess the diagnostic performance of a real time PCR assay for detection of Salmonella spp in feces of hospitalized hor ses, compared to bacteriological culture. The study approach involved several steps. First, to ensure that primers and probes used in the real time PCR assay were specific and did not cross react with other enteric or non enteric organisms, two sets of pri mers and probes were selected from published PCR studies after conducting a bioinformatics analysis. Thereafter, the real time PCR assay was validated using each set of selected primers and probes. Validation steps included optimization of the primer/probe concentrations, assessment of analytical sensitivity and specificity, and determination of the assay detection limit using Phosphate b uffered s aline (PBS) and fecal samples spiked with Salmonella organisms. Analytical sensitivity of three commercially ava ilable DNA extraction kits was evaluated to determine the most optimal kit which was later used to extract DNA from the study samples. To assess the diagnostic sensitivity (detection of true positives) of the real time PCR assay, fecal samples from horses with signs of GI disease (including colic, diarrhea, or fever and leu k openia) that were classified as Salmonella culture positive on at least one fecal samples were used in the analysis. To assess the diagnostic specificity (detection of true negatives), fecal samples from two different groups of horses were used in the analysis. The first group included fecal samples from horses without signs of GI disease that were classified as Salmonella culture negative on five fecal samples. The second group included fecal samples from horses with signs of GI disease that were classified as Salmonella culture negative on at least three fecal

PAGE 19

19 samples. A real time PCR assay with a for detection of Salmonella spp in fecal specimens of study horses w as considered an appropriate diagnostic tool Objective 4: Awareness and Relevance o f Hospital Surveillance and Infection Control Services a mong Ref erral Veterinarians a nd Clients The objective of this study was to assess the awareness and relevance of a hospital surveillance and infection control program among referral veterinarians and clients who refer or send horses to a referral hospital for vete rinary care A n electronic and mail survey was conducted and a convenience sampling approach was used to select eligible RDVMs and clients. Two separate 2 page questionnaires were designed for RDVMs and clients. Initially an electronic (email) survey was s ent to all enrolled RDVMs (n=242) and clients (n=483) that had email addresses and a remainder email was sent 10 days later to urge the respondents to complete the survey. Data for the respondents were entered into an excel database for analysis. A mail su rvey was later mailed out to all enrolled RDVMs (n=225) and clients (n=2612) that did not have an email address but had a physical mailing address (and were never contacted during the time when the electronic survey was administered). One month was allowed for responses to be returned by mail. Data for the respondents were entered into an excel database for analysis.

PAGE 20

20 CHAPTER 2 LITERATURE REVIEW Etiology Salmonellosis is caused by a variety of strains of Salmonella species, all of which are Gram negative, facultative anaerobic, motile, non lactose fermenting, rod shaped bacteria that belong to the family Enterobacteriaceae. 14 All Salmonella species are classified as S. enterica or S. bongori. Salmonellae are subdivided into serogroups and then further into serotypes or serovars based on testing with specific antisera. 1 5 The most important serogroups in veterinary medicine are A, B, C, D, and E. 1 6 Within the species of Salmonella enterica over 2,000 different serotypes of veterinary and medical importance af fect the gastrointestinal tract and result in diarrhea alone, or diarrhea in conjunction with fever, anorexia, depression, and shock. 14 The most frequently isolated serotypes in horses include S. enterica serovar Typhimurium S. enterica ser. Anatum S. en terica ser. Newport S. enterica ser. Krefeld and S. enterica ser. Agona 1 7 The only host adapted serovars reported for horses is Salmonella sp. ser. Abortusequi, causing abortion between 5 and 10 months of gestation. 18 Pathogenesis Once salmonellae infect a host, the first line of defense encountered is the acid barrier of the stomach. 1 9 Organisms that survive the acid barrier travel to the small intestine, where they are exposed to secretory products of the intestine (such as IgA, defensins, bile salts, a nd intestinal mucus) and to intestinal microflora that prevent bacteria from penetrating enterocytes. 20 Environmental factors in the intestinal lumen (such as oxygen concentration, osmolarity, and pH) affect the expression of Salmonella

PAGE 21

21 invasion genes (whi ch determine the release of bacterial products required for invasion of host cells). 21 For successful progression of a systemic infection and colonization of the host after oral infection, penetration of the gut epithelium is an important initial step. M ic rofold cell (M cells) located in the follicle primary intestinal epithelial cell type targeted for invasion by Salmonella in the mouse 22 and in cattle, S Typhimurium invades both M cells and enterocytes with no predilection for a particular cell type. 2 3 In adult horses, Salmonella primarily infects the cecum and proximal colon, causing typhlocolitis and the ability to disseminate beyond the intestine and cause enteric fever is limited. 24 In foals, howev er, salmonellosis is often associated with bacteremia The ability of Salmonella to cause enterocolitis depends on the ability of the bacteria to invade the gastrointestinal mucosa. 25,26 After crossing the intestinal mucus layer, salmonellae interact with both enterocytes and M cells. 21 Once attached to the enterocytes or M cells, the organisms are rapidly internalized. Salmonellae have the ability to induce endocytocis (ruffling) in otherwise nonphagocytic cells and enterocytes. This process involves the f ormation of large membranes around the bacteria by the host cell (enterocytes and M cells), as well as cytoskeletal rearrangements within the host cell itself. Once inside the cells, salmonellae migrate toward the submucosa of the small intestine where the y interact with macrophages and lymphocytes. 2 7 Salmonellae have the ability to survive within macrophages and use them to spread beyond the intestine. 1 9 They are transported via the lymphatics to draining lymphoid nodes and then to the thoracic duct, where they enter the systemic circulation. During the asymptomatic phase of the infection, the organisms are localized to the

PAGE 22

22 intestine and replicate within macrophages and epithelial cells. After becoming enveloped by macrophages, vacuoles called phago lyso some s are formed from the fusion of the ends of membrane ruffles and lysosomes contained within the macrophage. When a critical number of organisms have replicated, clinical signs result from the secretion of cytokines by the infected cells. 2 8 A characteristic feature of Salmonella infection is the induction of an early inflammatory response in the intestinal epithelium, resulting in the infiltration of polymorphonuclear leukocytes. The induction of such a response is likely due to the production of cytokines o r other proinflammatory molecules by natural killer cells and macrophages. 28 The inflammatory response contributes to the pathophysiology of the infection, characterized by inflammatory diarrhea. 2 7 Diarrhea associated with salmonellosis Diarrhea associate d with salmonellosis has multiple causes. 24 Salmonella cytotoxins inhibit protein synthesis in mucosal cells, causing morphologic damage and altered permeability. 2 9 Virulent Salmonella also produce an enterotoxin similar to the heat labile toxin (LT) produ ced by Escherichia coli. 30,31 The enterotoxin contributes to but is not required in the pathogenesis of diarrhea. 32,33 Salmonella enterotoxin increases secretion of chloride and water by colonic mucosal cells in many species, including horses, by increasin g intracellular cyclic adenosine monophosphate concentrations. The ability of virulent salmonellae to cause diarrhea appears to be associated most closely with the ability to invade enterocytes and to trigger an inflammatory reaction in intestinal tissue. 2 6,34 Effector proteins (SopB) injected into enterocyte cytosol by the type III secretory system of invading salmonellae stimulate chloride and fluid

PAGE 23

23 secretion. 35 Salmonella invasion of enterocytes is also a potent activator of inflammatory chemokine and cy tokine production, resulting in the recruitment of leukocytes, particularly neutrophils, and activation of resident macrophages and mast cells. Products of these activated leukocytes, including prostaglandins, leukotrienes, reactive oxygen metabolites, and histamine, are potent stimulators of chloride secretion in the colon of many species. 25,36 38 The enteric nervous system integrates the diverse processes of pathogen recognition, triggering of the inflammatory response, and induction of enterocyte fluid s ecretion. 3 8 Many of the inflammatory mediators stimulate colonic secretion by prostaglandin dependent mechanisms, resulting in increased intracellular cyclic adenosine monophosphate or calcium concentrations or both in mucosal cells. 36 In addition, these m ediators and the enteric nervous system may stimulate secretion by p rostaglandin independent mechanisms, inhibit sodium and water absorption, cause motility disturbances, and potentiate tissue injury, all of which enhance the pathogenicity and disseminatio n of Salmonella and contribute to the pathogenesis of diarrhea. 36,38 Neutrophils recruited to the mucosa by signals generated by the infected enterocytes physically contribute to mucosal injury by producing a variety of products that are lethal to pathogen s but are also toxic to host cells. 39 40 Moreover, neutrophils attracted to infected epithelial cells accumulate beneath the monolayer, lifting it off the basement membrane in sheets. Neutrophils also migrate across the epithelial monolayer in potentially massive numbers. Transepithelial migration of neutrophils increases the permeability to macromolecules, bacterial products, and even bacteria. 40 Potentially massive losses of electrolytes, water, and protein can occur depending on bacterial and host factor s. Perhaps most devastatingly,

PAGE 24

24 mucosal injury and altered permeability allow systemic absorption of bacterial products and dissemination of bacteria, resulting in systemic inflammatory responses such as occur with endotoxemia and bacteremia 24 Salmonella V irulence Mechanisms During a long standing association with its hosts, Salmonella enterica has evolved a sophisticated virulence mechanism for interaction with the host and modulation of host cell functions. This virulence mechanism includes a large numbe r of genes that are necessary for successful pathogenesis of Salmonella infections. 41 Many of these virulence genes protect Salmonella against damage generated by defense mechanisms of the innate and adaptive immune system of the infected host, or allow Sa lmonella to cope with nutritional deprivation imposed by specific host environments. 42 Several of the Salmonella virulence genes are clustered in certain areas of the Salmonella be en described and two of them, SPI 1 and SPI 2, contain a large number of genes encoding type III secretion systems for virulence proteins. The type III secretion system is a pathway by which virulence proteins are transported from the bacterial cytoplasm t o the extracellular space. Type III secretion systems are structurally and functionally related to the flagella assembly systems, and typically contain more than 20 different protein subunits that are located in the inner and outer membrane as well as in t he periplasm and cytoplasm of the bacterial cell. 43 The ability to invade non phagocytic cells and the intracellular survival and replication of Salmonella in phagocytes of the host are two hallmarks of Salmonella pathogenesis. 42 Salmonella are the only s pecies that contain two type III secretion systems, which are encoded by two distinct gene clusters (SPI 1 and SPI 2). These two

PAGE 25

25 type III secretion systems play different roles during pathogenesis; with SPI 1 being required for initial penetration of the i ntestinal mucosa and SPI 2 necessary for subsequent systemic stages of infection. 43 The SPI 1 encoded type III secretion system translocates effector proteins into the cytosol of host cells. This system is required for invasion of non phagocytic host cells (including epithelial cells) and enteropathogenesis, 44,45 while the SPI 2 encoded type III secretion system is required for intracellular survival in macrophages. 22 The molecular function of type III secretion systems is secretion of substrate proteins w hich fall into different functional classes. 42 One subset of substrate proteins is required to allow the translocation of a second subset of substrate proteins into the eukaryotic target cell. This second subset of substrate proteins is referred to as effe ctor proteins. SPI 1 encoded effecto r proteins include SipA, SipB, SipC, SipD, SptP, SopA, and SopD, while SPI 2 encoded effector proteins include SseB and SpiC. Other effector proteins encoded outside the SPI locus include SopB, SopE, and SopE2. There is evidence for a third subset of substrate proteins that are not translocated but appear to have regulatory functions. 42 Expression of Salmonella virulence genes is regulated by factors imposed on the pathogen by host microenvironments. Environmental conditi ons that affect the ability of Salmonella to enter host cells include oxygen levels, osmolarity, bacterial growth state, and pH. 4 6 The actual mechanisms, how the environmental signals are sensed, and how they influence gene expression, are not fully elucid ated. 42 The effects of these environmental factors are mediated by regulation of expression of transcriptional

PAGE 26

26 regulator HilA. 4 7 HilA expression is regulated by genes encoded on SPI 1, SPI 4, and SPI 5. 4 8 Salmon ella Virulence Genes Required f or Invasion o f Intestinal Epithelial Cells During the intestinal phase of Salmonella infection, the SPI 1 encoded type III secreted proteins play a crucial role, affecting invasion of intestinal cells, induction of inflammatory cell recruitment, and fluid secretion. 45 Studies have demonstrated that disruption of the SPI 1 encoded type III protein secretion system decreases or abolishes the ability of S Typhimuirum to invade the intestinal epithelium, which correlates with its ability to elicit an inflammatory response a nd thereby induce diarrhea. 48,49 51 Upon contact with intestinal epithelial cells, S Typhimurium translocates bacterial effector proteins into the host cell cytosol via the SPI 1 encoded type III secretion system. Some of these proteins have kinase, phosph atase, or actin binding activity, and once in the epithelial cell cytosol, they alter host cell signaling pathways that promote changes in the cytoskeleton. Pseudopods or membrane ruffles are formed on the host epithelial cell membrane, with consequent bac terial internalization and changes in host cell gene expression. 52 The SPI 1 encoded effector protein, SipC, acts as translocase and translocates itself into the host cytosol via the SPI 1 encoded type III secretion system. This protein bundles actin filam ents and nucleates actin polymerization which results in cytoskeletal rearrangements within the infected epithelial cell. 52 SipA, another SPI 1 encoded type III secretion protein, is not required for invasion but inhibits depolymerization of actin filament s by directly binding to and thereby decreasing the critical concentration of actin. 53 Therefore, SipC is essential for actin nucleation and bundling of actin filaments whereas SipA acts by enhancing the efficiency of this process. 54

PAGE 27

27 Other effector protei ns encoded by genes outside the SPI 1 locus, including SopE and its homologue SopE2, play a role in recruiting the actin nucleating complex to the membrane ruffles, thereby contributing to the invasion phenotype. 5 5 SopE protein targets host cell signal tra nsduction pathways by acting as a nucleotide exchange factor for the GTPases CDC42 and Rac1. 5 6 The activation of CDC42 and Rac1 results in the reorganisation of the actin cytoskeleton, leading to the uptake of Salmonella SptP, is an SPI 1 encoded type II I secretion protein that acts by reversing the cytoskeletal changes induced by Salmonella during invasion, restoring the normal structure. 5 7 Though SptP disrupts the actin cytoskeleton of epithelial cells, it is not required for cell invasion. 5 8 However, S ptP is required to switch off actin reorganization during host cell invasion thereby contributing to termination of the invasion process. 5 7 The effect of SptP antagonizing the action of other bacterial effector proteins clearly indicates that Salmonella is able to finely regulate cellular pathways in favor of its own purposes. Bacterial uptake by epithelial cells is quickly followed by changes in host cell gene expression. Transcription activators for the production of pro inflammatory cytokines including IL 8 are induced that provoke inflammatory responses. 5 9 IL 8 is a chemo attractant for neutrophils. Infiltration of neutrophils into the lamina propria occurs shortly after infection with S Typhimurium, and is followed by massive migration of neutrophils t hrough the epithelium into the intestinal lumen. 23,60 Some studies have indicated that IL 8 recruits neutrophils to the subepithelial space rather than through transepithelial migration into the intestinal lumen. 61,62 Salmonella induced IL 8 secretion by e pithelial cells is dependent on the mitogen activated protein kinase pathway and activation of the

PAGE 28

28 transcription factor NF kB, and requires a functional SPI 1 encoded type III secretion system. The Salmonella proteins, SipA and SopA, are involved in induct ion of neutrophil transepithelial migration. These two SPI 1 encoded effector proteins are translocated into the cytosol of the host cell via the SPI 1 type III secretion system. 63 Depending on the species infected, Salmonella infection may or may not res ult in enteritis and diarrhea. In mouse, no diarrhea occurs and the response to infection is mostly with mononuclear infiltration in the intestine. 64 In calves, S Typhimurium infection results in enteritis in which neutrophils are the primary inflammatory cells involved. 64 A characteristic feature of Salmonella induced enteritis is an intense intestinal secretory and inflammatory response including the induction of polymorphonuclear cell transmigration through the intestinal epithelium in humans 6 5 The SPI 5 encoded gene, SopB, has been linked to the pathogenesis of diarrhea 66 ,6 7 and its expression is dependent on the regulator sirA, which is also an activator of the SPI 1 regulator hilA. 6 8 SopB has inositol phosphate phosphatase activity 6 9 and is not requir ed for invasion, but contributes to diarrheal symptoms by triggering loss of electrolytes and fluid secretion of epithelial cells of the intestine. 6 7 SopB is thought to mediate fluid secretion by increasing chloride secretion. 70 The Salmonella protein SopD which is also secreted in an SPI 1 dependent manner, has an additive effect to SopB in the induction of enteritis. 71 Another protein, SopA, influences the intestinal inflammatory response by a mechanism distinct from SopB and SopD. SopA is involved in in duction of neutrophils, a phenomenon that is not influenced by SopB or SopD. 72 The effector proteins SipA, SopA, SopB, SopD, and SopE2 have been demonstrated to act in concert to induce diarrhea. 73

PAGE 29

29 Survival o f Salmonella Following Intestinal Invasion Salmo nella is a facultative intracellular pathogen that can survive and replicate inside host phagocytic cells, most prominently in macrophages 74 and can utilize phagocytes as vehicles for dissemination to other tissues such as the spleen and liver. 7 5 The abil ity to grow within macrophages and polymorphonuclear phagocytes is a prerequisite for Salmonella virulence. 74,76 79 Once Salmonella have crossed the epithelial layer that separates the intestinal lumen from host tissue, the bacteria first encounter the cel 80 82 The cells encountered include resident phagocytes and dendritic cells. Macrophages in particular are crucial to the initial defense against Salmonella ; serving to ingest bacteria and g enerate inflammatory signals in response to infection. 83 85 After uptake by a phagocytic cell, Salmonella does not enter the intracellular cytoplasm but resides inside the phagosomal vacuole. In this environment, Salmonella has to be able to cope with envi ronmental changes such as rapid decrease in pH and nutritional deprivation that are restrictive for the growth of phagocytosed bacteria. In addition, phagosomes containing ingested particles usually enter the degradative pathway of the cell. After fusion w ith lysosomes to form a phagolysosome, phagocytosed bacteria will encounter toxic substances such as hydrolytic enzymes (proteases, lysozyme), small cationic proteins (i.e. defensins), and enzymes that produce reactive forms of oxygen to cause the so call ed oxidative burst (superoxide radical, hydroxyl radical), as well as reactive nitrogen intermediates. 42 The strategy by which Salmonella modifies the phagosome to survive and replicate is not yet fully understood. Suggested mechanisms include inhibition of the fusion of Salmonella containing vacuoles with lysosomes, 8 6 delay in acidification of the

PAGE 30

30 phagosome, 8 7 or the generation of Salmonella containing vacuoles with a composition different from that of normal phagocytes, 8 8 and avoidance of the NADPH oxida se dependent killing by macrophages. 8 9 An SPI 2 encoded substrate protein, SpiC, has been shown to interfere with normal cellular vesicle trafficking. 90 Therefore, SPI 2 may be involved in the altered characteristics of the Salmonella containing vesicles. The molecular functions of SPI 2 have not been characterized in as much detail as those of SPI 1. 42 Salmonella can induce death of host phagocytic cells. Salmonella induced macrophage cell death is largely due to expression of genes associated with invasi on. SipB protein has been identified as the bacterial protein responsible for induction of apoptosis. 91 SipB is translocated into the host cell cytosol via the SPI 1 encoded type III secretion system, where it binds to and activates caspase 1, an intrace llular cysteine prote ase also known as IL 1 cleaves and activates IL 91 Therefore, Salmonella induced macrophage apoptosis results in release of active IL Salmonella elicited in flammation. Multi Drug Resistant Salmonella Strains i n Horses Multi drug resistant (MDR) Salmonella infections in horses have been documented. In two previous outbreak investigations of nosocomial salmonellosis in horses, cattle, and alpa cas, multi drug re sistant strains of Salmonella Newport were identified as the causative strain. 5,92, In one study, 17 of 54 horses infected with the outbreak strain died during hospitalization (but it was not clear how many horses were affected with diarrhea). 5 In another study, 1 of 2 horses infected with the outbreak strain developed enterocolitis, diarrhea, septicemia and subsequently died. 92 Several

PAGE 31

31 Studies in humans have demonstrated that infections with drug resistant non typhoid Salmonella serotypes are associated w ith excess morbidity and mortality. 93 100 Multi drug resistance (MDR) typically occurs as a result of accumulation of multiple mutations and/or the horizontal transfer of resistance genes on mobile genetic elements such as plasmids and transposons along wi th novel genetic elements such as integrons and resistance islands. 101 104 Most of the antimicrobial drug resistance determinants of Salmonella are encoded on plasmids and on the multidrug resistance region of SPI 1. 10 5 Plasmids are small, circular, self replicating DNA elements that are capable of transfer via conjugation. 103,106 Resistance genes encoded in plasmids are often located within genetic elements called transposons. These elements possess the transposase function that enables the transposon to recombine into the bacterial chromosome or plasmids. 103 Horizontal transfer of plasmids to other cells often provides a selective advantage to the organism if the plasmid carries genes encoding antimicrobial resistance, persistence, environmental adaptabil ity, heavy metal resistance, and/or virulence. 103,107 109 Plasmids are classified by incompatibility (Inc) groups, which are named as such because two members of the same Inc group cannot be stably maintained in a bacterium during cell division. 1 10 In Sal monella isolated in the United States, IncA/C plasmids have been associated with an expansion of MDR in several serovars of Salmonella These plasmids are often large (150 200 kb) and some can carry genes encoding for resistance to 10 or more antimicrobial s. 1 11 Multi drug efflux pumps of the resistance nodulation division (RND) family encoded on SPI 1 play a major role in providing both intrinsic and elevated levels of

PAGE 32

32 resistance to a wide range of noxious compounds in Gram negative bacteria. 1 12 ,1 13 In Salm onella increased expression of the multi drug AcrAB TolC efflux pump has been shown to be a significant determinant of microbial resistance. 1 14 This system confers innate multiple antibiotic resistance. 11 5 Disruption of the multi drug AcrAB TolC efflux pu mp has been shown to result in decreased pathogenicity associated with reduced expression of genes located on SPI 1 and other attributes required for anaerobic growth, motility, and host cell invasion. 11 6 The mobility of SPI 1 coupled with the ability of v arious antibiotic resistance genes to be integrated and lost from the chromosomal resistance locus allows for the transfer of stable antibiotic resistance to most of the commonly used antibiotics and adaptation to new antibiotic challenges. 114 A class of m obile genetic elements, the integrons, are considered efficient means by which transfer of resistance markers among unrelated bacterial populations can be facilitated as part of a transposon of the Tn 21 family or independently on broad host range plasmids. 114 ,11 7 Integrons contain a recently described system for site specific recombination, similar to that in temperate bacteriophages. 11 8 These structures are naturally occurring gene expression systems that can potentially capture and integrate one or more g ene cassettes and convert them into functionally expressed genes. It is these gene cassettes that encode open reading frames (ORFs) corresponding to the resistance determinants of several antimicrobial agents. MDR encoding integrons are widespread in S Ty phimurium DT104, other phage types of S Typhimurium as well as other serovars of Salmonella and many other Gram negative bacteria. 106 ,11 9 121 Host Immunity against Salmonella Infections Host factors that restrict gastrointestinal colonization and invasion by pathogens include gastric pH, commensal gastrointestinal flora, gastrointestinal motility, the

PAGE 33

33 mucosal barrier and mucosal immunity. 24,25,122 Gastric acidity is an important defense mechanism preventing live organisms from reaching the intestine. Alter ing the gastric pH with histamine receptor antagonists, for example, may increase susceptibility to infection. Gastrointestinal flora inhibits the proliferation and colonization of Salmonella by secreting bacteriocins, short chain fatty acids, and other su bstances that are toxic to Salmonella In addition, elements of the normal flora compete for nutrients and space, especially on the mucosa. 25 Being predominantly anaerobic, the normal flora maintain a low oxidation reduction potential in the environment of the large intestine, which inhibits the growth of many bacterial pathogens. 123 The importance of normal host gastrointestinal ecology is illustrated by the fact that disturbances of the colonic flora with antibiotics, changes in feed, ileus, or other unde rlying gastrointestinal disease greatly increase the susceptibility of the host to infection by Salmonella often resulting in serious disease. After surviving these host factors, Salmonella encounters other factors at the intestinal mucosal surface includ ing mucus which forms a physical barrier, lysozyme, lactoferrin, and lactoperoxidase. 1 24 The immune status of the host may be one of the most important factors determining not only the susceptibility to Salmonella infections but also the degree of invasio n and subsequent outcome of the infection. 24 During the initial stages of Salmonella cells, natural killer T cells, neutrophils and macrophages. 12 5 This first line of defense invol ves production of high levels of gamma interferon (IFN alpha (TNF derived cytokines act by enhancing the bactericidal capacity of phagocytes, facilitatin g

PAGE 34

34 antigen presentation, and influencing the T helper cell polarization of the immune response. 12 5 Opsonizing antibodies and activation of the complement cascade are important in fighting systemic invasion by Salmonella by increasing the efficiency of phago cytosis and by direct bactericidal activity. 24 The adaptive immune response with both humoral and cell mediated immune responses, is involved in the acquired resistance to Salmonella infection. 1 7 The humoral response involves production of IgA by plasma ce lls. IgA is the principal antibody isotype involved in mucosal immunity, and acts by binding to surface antigens and preventing attachment and penetration of salmonellae. 20 Cell mediated immunity is activated by the different antigens expressed in S almonel la which induce a specific T helper cell response. On the basis of differential cytokine expression, both T helper 1 (Th1) and T helper 2 (Th2) type responses can be identified in most cases of infection by facultative intracellular pathogens in mice 2 8 T h1 cells produce cytokines such as 12), which regulate cell mediated protective immune responses against intracellular organisms. In contrast, Th2 cells produce cytokines such IL 4, IL 5, IL 10 and IL 13 which re gulate humoral immune responses which may not be protective. 2 8 ,12 6 Salmonella are intracellular bacteria which can survive in macrophages and dendritic cells. 2 7 Consistent with the notion that Th1/Th2 ratio predicts the competence of the response to intrac ellular pathogens, there is some early evidence that invasive Salmonella generally elicit a Th1 immune response; however, there are also reports describing both Th1 and Th2 responses to Salmonella in mice with the predominant response reflecting the parti cular pathway of antigen processing (MHC II versus MHC I

PAGE 35

35 dependent) in macrophages and dendritic cells. 2 8 Another adaptive immune response in the early stages of infection is cell lymphocytes, which function as effector cells, and cells in the late stages of disease in mice 127 Epidemiology and Infection Control Aspects of Nosocomial Salmonella Infections in Hospitalized Horses Over the past thirty years, outbreaks of nosocomial Salmonel la infections in hospitalized horses have been reported, with varying consequences. In 1981 1982, the University of California Davis, Veterinary Medicine Teaching Hospital experienced an outbreak of nosocomial salmonellosis due to Salmonella enterica serot ype Saintpaul. 1 The outbreak extended for 10 months, resulting in severe disruption of hospital routine and temporary closure of the hospital for 3 months to facilitate remodeling and disinfection. In 1995, an outbreak of nosocomial salmonellosis due to S. Infantis at Colorado State University resulted in temporary closure of the hospital for 3 months and an estimated $500,000 in lost revenues and facility renovation. 2, 128 In 1996, an outbreak of equine salmonellosis due to S. Typhimurium occurred at the Mi chigan State University. 3 Unique features of this outbreak included a high case fatality, zoonotic infection, and closure of the hospital for 1 month for complete disinfection. In 2000, an outbreak due to a multi drug resistant strain of S. Typhimurium occ urred at Purdue University, 4 the hospital was closed for 3 months. In this outbreak, estimated losses included equine mortality, $250,000 to $300,000 for cleaning and disinfection, reduced caseload, and impaired student education. In 2004, an outbreak of e quine salmonellosis due to S. Newport occurred at the University of Pennsylvania veterinary teaching hospital. 129,130 The hospital was closed for 3 months, and students had to go out of state

PAGE 36

36 for large animal clinical training. Estimated losses during this outbreak included equine mortality and over $4 million for hospital renovations, cleaning and disinfection procedures, as well as a reduced caseload for 10 months following re opening. The most recent reported outbreak of salmonellosis was associated with S. Newport at Colorado State University veterinary teaching hospital in 2006 92 In this outbreak, a total of 8 animals (4 alpacas, 2 horses, 1 goat, 1 cow) were found to be shedding the outbreak strain, all but 1 shed Salmonella in the absence of clinical signs or before the onset of disease. Unlike previous outbreaks reported above, in this outbreak the hospital was not closed. The threat of closure was avoided primarily because of aggressive surveillance and mitigation strategies. 92 Surveillance a nd In fection Control Control and prevention of outbreaks of nosocomial Salmonella infections in hospitalized horses requires careful formulation, implementation, and evaluation of a hospital surveillance and infection control program. For an early detection sys tem, the monitoring of nosocomial Salmonella infections in hospitalized horses is fundamental. From a hospital risk management point of view, a working case definition of nosocomial infection must be formulated, even when Salmonella serotyping data are not readily available. Case definition for nosocomial Salmonella infections In general, salmonellosis is suspected to be of nosocomial origin when an infection is identified after animals have been hospitalized for 72 hours or longer, and when the serotype an d antimicrobial susceptibility pattern of Salmonella isolates from the primary and nosocomial cases are the same. However, in published epidemiologic studies, the criteria used to define nosocomial Salmonella infections in horses have varied. For

PAGE 37

37 example, in one study, an outbreak was considered nosocomial in origin because a relatively large number of cases shedding the same serotype were detected within a short period of time, and the serotype involved had seldom previously been isolated from hospitalized animals at that hospital. 1 In another study, infection with S. Krefeld or S. Typhimurium was considered nosocomial when the mean time from admission to 1 31 In a recent study, 1 32 horses were classified as nosocomial Salmonella cases using the following criteria: Nosocomial cases were horses that tested negative for Salmonella in samples obtained at the time of admission and tested positive in samples collected 48 hours after admission, and the primary case or source of nosocomial in fection was an index case that had positive results for Salmonella in a sample collected at the time of admission, and shared the same serotype and antibiogram as the isolate from the nosocomial case, and there was an overlap between admission and discharg e dates of the primary and nosocomial cases. Another source of nosocomial infection was environmental contamination. Horses that had negative results for Salmonella in samples obtained at the time of admission but had positive results thereafter and shared the same serotype and antibiogram as a Salmonella positive environmental sample collected during the period of hospitalization were also classified as nosocomial cases. When environmental contamination was attributed as the source of infection, the nosoco mial case was never exposed to the primary case associated with the environmental contamination (i.e. there was no overlap between admission and discharge dates of the primary and nosocomial cases because the primary case had

PAGE 38

38 already been discharged from t he hospital or euthanized before the nosocomial case was admitted). In addition to phenotyping methods such as serotyping and antimicrobial susceptibility testing, genotyping methods such as pulsed field gel electrophoresis (PFGE), plasmid profile analysis and phage typing have been used to determine if Salmonella isolates are genetically related, providing further evidence of nosocomial infection. 3,4,133,134 In a study that followed an outbreak of equine salmonellosis, presence of plasmids, examination o f antibiotic profiles, and production of bacteriocins and hemagglutination of erythrocytes were used to indicate that equine and environmental isolates had a common origin. 1 33 In another study, PFGE was used in addition to serotyping and antimicrobial susc eptibility testing, to characterize the outbreak strain and isolates were considered nosocomial if the PFGE pattern was similar to that of the outbreak strain. 3 Another outbreak investigation used PFGE and phage typing to identify the outbreak strain, in a ddition to serotyping and antimicrobial susceptibility testing. 4 Finally, in a study that examined the epidemiological relationship between Salmonella cases by comparing isolates from a veterinary hospital over a period of six years, PFGE, presence of the insertion element IS 200 plasmid profiles, and antimicrobial resistance patterns were used to show that isolates from the veterinary hospital originated from a common source. 1 34 From a hospital risk management point of view, a working case definition must be formulated even when Salmonella serotyping data are not readily available. At the University of Florida: Large Animal Hospital (UF LAH), the serotyping of Salmonella isolates is performed at the US Department of Agriculture: National Veterinary Services

PAGE 39

39 Laboratory in Ames, Iowa, and it may take 4 weeks or longer to obtain the serotyping results. Thus, it becomes imperative to have a working case definition that excludes serotyping data but permits early detection of nosocomial Salmonella infections, so t hat enhanced infection control measures can be implemented to reduce the risk of disease transmission. At the UF LAH, the following two working case definitions are used based on two potential sources of nosocomial infection: another in patient (index case ) or environmental contamination (Figure 2 1). First, if the source of nosocomial infection is another patient, then a horse will be considered a suspect for nosocomial Salmonella infection or colonization when: (i) it tests positive on a fecal sample that is collected 48 hours after admission; (ii) the serogroup of the isolate from the suspect nosocomial case is the same as that of the isolate from the suspect primary case; and (iii) the antimicrobial susceptibility pattern of the isolate from the suspec t nosocomial case is similar to that of the isolate from the suspect primary case. In addition, the following two parameters are evaluated: (iv) if there is a temporal overlap of admission and discharge dates of the suspected nosocomial and primary cases; and (v) there is a spatial overlap of housing location for the suspected nosocomial and primary cases (e.g., if both the suspected nosocomial and primary cases are housed in the same barn). The first four conditions above (i, ii, iii, iv) are required, bu t the last one (v) may or may not affect the decision to classify a horse as a nosocomial case. Second, if environmental contamination is considered the source of infection, then a horse will be considered a suspect for nosocomial Salmonella infection when : (i) it tests positive on a fecal sample that is collected 48 hours after admission; (ii) the serogroup of the isolate from the suspect nosocomial case is the same as that of the environmental isolate; (iii) the

PAGE 40

40 antimicrobial susceptibility pattern of the isolate from the suspect nosocomial case is the same as t hat of the environmental isolate; and (iv) there is no time overlap between admission and discharge dates of the suspect nosocomial and primary cases; however, the Salmonella isolate recovered from the suspect primary case shares the same serogroup and ant imicrobial susceptibility pattern as a Salmonella positive environmental sample collected during the period of hospitalization. These two working definitions of nosocomial Salmonella infection have served well at the UF LAH over the last 10 years to identi fy nosocomial Salmonella infections, and to justify the need to implement enhanced infection control measures to reduce the risk of disease transmission in hospitalized horses. A retrospective analysis using Salmonella or longer after each suspect nosocomial case was identified, revealed that the frequency of misclassification of nosocomial cases was ~ 1%. 135 Environmental contamination as a source of nosocomial Salmonella infections in hospitalized horses The environme nt can be an important source and reservoir for exposure and transmission of Salmonella in veterinary hospitals. In several reports of nosocomial outbreaks of salmonellosis, environmental contamination was implicated in the spread of S almonella among patie nts and in the persistence of salmonellae in the hospital environment. 2,4,133,134 In one study, persistence of S. Typhimurium in the environment was identified as the source of nosocomial infection for several horses. 3 In that study, S. Typhimurium was iso lated from hospital personnel, shared equipment, and stalls. A hospitalized foal was identified as the point source of infection. To confirm if the environment was a source of nosocomial infection, serotype information, antimicrobial

PAGE 41

41 susceptibility, and PF GE patterns were used to compare the environmental isolates and the isolate recovered from the point source foal. In another study, environmental contamination was suggested to be the source of infection for other horses during an outbreak of salmonellosis due to S. Typhimurium in a university hospital. 4 The point source of environmental contamination was a horse that presented with colic. Salmonella Typhimurium was isolated from stall drains, surgery pads, forklift tires, and the ambulatory garage floor. S imilarities of isolates based on serotyping, antibiogram, phagetyping and PFGE patterns were used to indicate that a common source strain of S. Typhimurium was responsible for environmental contamination. In a different study, environmental contamination c ontributed to the wide spread nature of infection during an outbreak of S. Infantis in horses and food animals. 2 Salmonella Infantis was isolated and mats in stalls a nd recovery rooms; the original source of S. Infantis was not determined. Hospital environment can serve as a reservoir, leading to persistence of salmonellae in hospital environments over protracted periods. In a study that followed an outbreak of salmon ellosis due to S. Infantis at a veterinary hospital, isolates obtained from patients or the hospital environment over a period of 9 years were compared using antimicrobial drug susceptibility patterns, PFGE, and presence of integrons to determine whether i solates were epidemiologically related. 128 Results of the genetic analysis in combination with clear epidemiological links indicated that environmental contamination arising from the outbreak persisted across years despite rigorous hygiene and biosecurity precautions and may have led to subsequent nosocomial infections. In

PAGE 42

42 another study at a veterinary hospital, 28 isolates of multi drug resistant (MDR) S Heidelberg obtained from horses over a period of six years were compared using PFGE, IS 200 element pro files, antimicrobial resistance patterns, plasmid profiles, and phage typing. 1 34 The PFGE patterns indicated that the MDR isolates from the veterinary hospital originated from a common source. The isolates also had indistinguishable IS200 profiles, the sam e antimicrobial resistance pattern and had at least one plasmid in common. In that study, no environmental isolates were tested. The isolation of a single strain of MDR S. Heidelberg from horses admitted to that hospital during the six year period is likel y explained by persistence of this strain in the hospital environment. Sampling In most veterinary hospitals with surveillance programs in place, horses known to be at a high risk of shedding Salmonella at admission and during hospitalization are targeted for routine sampling. This sub population includes horses that present with signs of gastrointestinal tract disease such as colic, horses with diarrhea, and horses with fever and leucopenia. At Colorado State University, fecal samples are obtained from all bovine inpatients for Salmonella culture, as well as from all equine colic patients at arrival and every other day after that. 1 3 6 At Michigan State University, fecal samples are collected on the day of admission and at various times thereafter from all ho rses with evidence of gastrointestinal tract abnormalities for diagnosis of Salmonella using bacteriological culture. 1 3 7 samples are collected from horses without clinical signs of gastrointestin al tract disease that are considered to be at risk for shedding salmonellae such as neonatal foals with systemic disease, mare and foal pairs when only one of the pair has diarrhea, and horses treated with antimicrobials for long periods. At Purdue Univers ity, fecal samples

PAGE 43

43 are collected from horses that present with diarrhea, or that develop diarrhea with leucopenia or fever after admission to test for Salmonella using bacteriological culture. 4 Several university veterinary hospitals have environmental sam pling as part of their surveillance and infection control programs. At Colorado State University, stalls in which Salmonella positive animals are housed are cultured before being made available for use with other patients. 128 At Michigan State University, stalls used for horses with diarrhea or that shed salmonellae in their feces are sampled and tested using bacteriological culture. 1 3 7 Environmental samples are also collected from other hospital areas that are considered at risk for Salmonella contaminati on such as surgery rooms, anesthesia induction and recovery rooms. Laboratory diagnosis Detection of Salmonella spp using bacteriological culture Most veterinary hospitals with a surveillance and infection control program use bacteriological culture for d etection of Salmonella spp. colonization or infection in horses, as well as environmental contamination of hospital facilities. While bacteriologic culture is the most common diagnostic procedure used for identification of horses infected with Salmonella culture procedures are not standardized among veterinary microbiology laboratories. 1 3 8 For example, in a recent study, 1 32 1 to 2 g of fresh feces was placed in 10 mL of selenite cystine broth (ratio, 1:10 to 1:5), and the broth was incubated at 37 o C overni ght for selective enrichment of Salmonella This laboratory procedure differs from that reported in another study 1 31 in which investigators placed 5 g of fresh feces in 100 mL of selenite broth (ratio, 1:20) and incubated the broth at 37 o C overnight. These differences can affect the epidemiologic sensitivity (i.e., false negatives) for diagnosis of Salmonella For example, in one study, weight of the fecal sample obtained from pigs had an effect on

PAGE 44

44 detection of Salmonella ; sensitivity increased from 32% to 63% when weight of the fecal sample increased from 1 to 10 g (diluted 1:9 [wt:wt] with buffer peptone water solution). 13 9 It is difficult to know how those results in swine can be extrapolated to horses. It is particularly difficult to assess how much misc lassification of infected or non infected horses may have occurred in previous epidemiologic studies. The study in swine used only 1 sample per pig. In the equine study that used 1 to 2 g of fresh feces, 1 32 the median number of samples collected was 3 for control horses and 4 for nosocomial case horses. Assuming the concentration of Salmonella in the first fecal sample collected in that study population was similar to that in swine, 13 9 it is possible the surveillance system may have missed Salmonella shedde rs at admission because of low sensitivity. Although bacteriological culture is considered the gold standard for detection of Salmonella spp in fecal specimens in horses, 6,8,10,92 it has some limitations. One limitation of using bacteriological culture t o monitor community acquired (infection contracted outside a hospital or infection present on admission) and nosocomial Salmonella infections in horses is the time required (3 to 5 days) to obtain positive laboratory results. In the absence of overt clinic al signs (diarrhea or fever and leucopenia), this prolonged detection time creates a delay in implementation of appropriate infection control measures such as isolation which are necessary to minimize the risk of nosocomial infections and environmental con tamination. In addition, serotyping of Salmonella isolates is often conducted at a reference laboratory (USDA National Veterinary Services Laboratory in Ames, IA), and it takes 2 weeks or longer to obtain the results.

PAGE 45

45 A second limitation of bacteriologica l culture is suboptimal sensitivity. 8 In a previous study that compared the sensitivity of bacteriological culture and PCR, all PCR positive horses were detected after a total of 3 samples/horse were submitted, whereas as many as 5 samples/horse were requi red to identify all culture positive horses. 1 1 In that study, 71 (65%) of 110 hospitalized horses were positive by PCR, compared with 11 (10%) horses that were positive by culture. Fifty six of 71 (79%) horses that were test positive by PCR were identified as test positive after the first sample from each study horse was submitted 63/71 (89%) were test positive by PCR after testing of the second sample, and all (71/71;100%) were identified as test positive after testing of the third sample. Seven of 11 (64 %) culture positive horses were identified after testing of the first sample for each study horse, 8/11 (73%) were identified as culture positive after testing of the second sample, and 9/11 (82%) were identified as culture positive after testing of the th ird sample. B acteriological culture of multiple fecal samples from horses has been shown to yield a greater proportion of Salmonella positive results compared to single fecal samples. 1 40 A requirement of multiple samples edding status is not best suited for hospital infection control programs because of the lag in time between identification of Salmonella positive horses and isolation from the general hospital population, and the cost of performing multiple cultures. 8 The prolonged time required for definitive identification of Salmonella spp and the suboptimal sensitivity of bacteriological culture underscore the need for a more rapid and sensitive method for detection of Salmonella spp in hospital surveillance programs.

PAGE 46

46 D etection of Salmonella spp using PCR Several PCR protocols have been developed and evaluated for potential use in hospital surveillance programs for early detection of Salmonella in hospitalized horses. The advantage of using PCR protocols to monitor Salm onella shedding in feces of hospitalized horses is that test results can 5 days), thus allowing an earlier implementation of preventative measures such as isolation and barrier nu rsing and, therefore, minimizing the risk of a potential outbreak of nosocomial infection. 7,8 Several studies (Table 2 1) have reported different estimates of sensitivity and specificity of real time PCR protocols for detection of Salmonella spp in fecal s amples of horses. These studies 9,10,12,13 concluded that PCR is a rapid, sensitive, and specific assay for detection of Salmonella spp that can be an alternative to conventional culture methods for surveillance, and can reduce the risk of nosocomial infect ions through the provision of highly accurate and rapid pathogen detection. Despite the availability of rapid and highly sensitive real time PCR tests for detection of Salmonella its acceptance as a surveillance tool has been very limited. One of the lim itations to the acceptance of PCR testing is that sampling methods used to establish the gold standard for true positives and true negatives have been non objective 10 13 and t his can le a d to misclassification bias and inconclusive results These sampling l imitations are showed in two previous real time PCR studies that targeted the invA gene. In the first study, a specificity estimate of 339/345 or 98% was reported for PCR when compared to culture. 13 However, in that study, fecal samples used to estimate sp ecificity were collected from horses and the environment of pens of a horse feedlot and the clinical status of study horses was not reported. Therefore it is not clear

PAGE 47

47 if horses in that study had colic with or without clinical signs of salmonellosis In ad dition, only one fecal sample was collected from each study horse. In a second study that also compared the diagnostic performance of PCR to culture, specificity estimates reported were 904/905 or 99% when fecal samples without enrichment were used and 889 /905 or 98% when fecal samples with enrichment were used. 10 In that study, horses used to assess specificity included horses with or without colic, and only one fecal sample was tested from each horses. It is possible that some of these horses with colic w ere infected with Salmonella and could not be detected based on one culture result only. In these two previous studies, 10,13 the lack of a gold standard definition for true negatives and the use of horses that tested culture negative on one fecal sample o nly may have led to misclassification bias which produces inconclusive results. Additional previous studies that compared the diagnostic performance of conventional PCR with culture identified more limitations on the proposed use of PCR as a surveillance tool for Salmonella in hospital infection control programs. An early study compared the diagnostic performance of PCR and culture for detection of Salmonella spp in fecal samples collected from horses with and without clinical signs of salmonellosis 11 The PCR protocol used primers for the highly conserved segment of the histidine transport operon gene of S. Typhimurium. Overall, more fecal samples were classified as positive using PCR, compared to culture. Among equine outpatients without clinical signs of salmonellosis, 26/152 (17%) tested positive by PCR and 0/152 (0%) tested positive by culture. In addition, among hospitalized horses, 71/110 (65%) tested positive by PCR and 11/110 (10%) tested positive by culture. This study revealed that PCR can yield m any false positive results (especially among equine outpatients

PAGE 48

48 without clinical signs of salmonellosis). In a hospital setting, the consequences of false positive results are significant because equine in patients with a positive result must be placed in isolation stalls, increasing the cost of hospital fees. Another study 8 compared the diagnostic performance of PCR and culture for detection of Salmonella spp in equine feces. The PCR protocol used the same primers as those by Cohen et al. (1996). 1 1 Results from this study were very informational because the analysis was were collected and cultured for Salmonella spp; the median number of samples collected per horse was 8 A total of 87 (75%) horses had 1 or more samples classified as positive by PCR versus 11 (9.5%) by culture. In this study, if culture were used as the gold standard, the specificity of the PCR protocol used would be very low. Among the 105 horses that we re culture neg negative for all samples tested. The authors concluded that the reasons why some fecal samples (from which Salmonella spp cannot be cultured) are PCR positive need to be determined before PCR can be incorporated into Salmonella surveillance programs for hospitalized horses. Another potential limitation of using PCR as a surveillance tool is that Salmonella can lose the target genomic sequence in a PCR assay, resulting in false negative results. 1 2,13 7 Environmenta l isolates of S Senftenberg and S Litchfield which carry a deletion encompassing a vast segment of the centisome 63 region of the Salmonella chromosome have been reported. 1 4 1 This deletion includes the entire inv, spa, and hil loci, which are required fo r entry of Salmonella spp into mammalian cells. Deletion of

PAGE 49

49 genes that are targeted in PCR assays creates ambiguity in the interpretation of PCR false negative results. An important consideration that should be take n into account before selecting PCR as a surveillance tool is that antibiogram and serotype data can only be obtained if PCR is used in combination with bacteriological culture. If an enrichment step is used prior to DNA extraction and PCR amplification, part of the enrichment broth can be stored and later tested using bacteriological culture followed by antimicrobial susceptibility testing and serotyping of samples that test po sitive by PCR. Antibiogram and serotyping data are needed to establish if there is an epidemiological relationship betwee n Salmonella cases or isolates (from potential primary and nosocomial cases), thus providing evidence that Salmonella infection or colonization is nosocomial in origin. In addition, this information can facilitate therapeutic decisions by clinicians in cas es where antimicrobial therapy is deemed necessary. Use of real time PCR protocols has the potential to improve the diagnostic performance of conventional PCR protocols used for diagnosis of Salmonella spp in equine feces. The use of a sequence specific f luorogenic probe in addition to conventional PCR primers allows for greater specificity for detecting a target sequence. 14 2 In addition, the fluorogenic nature of the probe allows for rapid detection, because a fluorescent signal can be detected as the rea ction progresses, negating the need for post PCR assay sample processing. Another advantage of real time PCR over conventional PCR is that it can minimize post amplification cross contamination or d tube systems are used. 1 0 ,13

PAGE 50

50 Enrichment of fecal specimen prior to PCR amplification can improve sensitivity of a PCR assay for diagnosis of Salmonella spp in equine feces. In one study in which enrichment was used prior to PCR amplification, PCR detecte d 10 0 colony forming units (CFU) of S. Enteritidis/g of feces, 14 3 and in another study, performed by the same author, where enrichment was not used, PCR detected 10 3 10 4 CFU of Salmonella spp/g of feces. 7 The PCR method that used enrichment was more sensi tive than the PCR technique that did not involve use of enrichment. 14 3 In another study where real time PCR was used to detect Salmonella spp in 911 fecal and enriched broth samples, three of the 911 (0.3%) fecal samples tested positive while 22 (2.4%) enr iched broth samples tested positive. 1 0 The type of DNA extraction method used can affect sensitivity, specificity, level of cross contamination, cost, labor, and time required for a PCR assay. In a previous PCR study that compared the relative sensitivity and specificity of three commercially available DNA extraction kits, three different sensitivities and specificities were reported: 100 and 98.1%; 80 and 100%; and 93.3 and 100%. 1 2 The kit with a sensitivity and specificity of 100 and 98.1% was considered optimal and was selected for use in that PCR assay. In another study, a DNA extraction method was reported to reduce chances of cross contamination during extraction and to be less expensive and required less labor and time compared to DNA extraction metho ds used in other PCR assays. 7,14 3 ,14 4 In addition, this method of DNA extraction did not require the use of potentially noxious organic compounds (e.g., phenol, chloroform, and isoamyl alcohol). 14 3 Risk management Several studies have demonstrated that iso lation, cleaning and disinfection, and traffic control can be effective in the control of Salmonella outbreaks in

PAGE 51

51 hospitals. 2, 3 ,14 5 ,14 6 The most commonly used protocol has been isolation of horses that pose a risk to other horses; these include horses with clinical signs of salmonellosis (diarrhea, fever and leucopenia), and horses that are suspected or confirmed to be Salmonella positive. Most veterinary hospitals maintain isolation units for this purpose, and horses are considered infectious and contagiou s until proven otherwise. A number of methods are employed to prevent and control microbial contamination during isolation, including the use of barrier nursing precautions such as examination gloves, protective coveralls or gowns, disposable boots when ha ndling infected horses, and foot baths or foot mats. Foot baths and foot mats have been shown to be effective in reducing bacterial contamination in veterinary hospital environments when used properly. 14 7 ,14 8 In addition to barrier nursing precautions, pr omoting personal hygiene habits among hospital personnel can raise awareness of the importance of such behaviors in controlling nosocomial disease. 14 6 These behaviors include washing hands or using alcohol based hand wipes or lotions between patients, clea ning boots, avoiding walking wounds or soiled bandages, cleaning up manure promptly, and not bringing personal pets into barn areas. One of the most important aspects of per sonal hygiene is washing operated sinks and soap dispensers or alcohol based hand disinfectant solutions should be available at key locations in the hospital (such as i solation facilities). Effective cleaning and disinfection of contaminated environments is one of the most important measures in preventing and controlling the spread of Salmonella in

PAGE 52

52 veterinary hospitals. Thorough cleaning of areas with fecal contamination such as stalls, water buckets or automatic watering apparatuses and drains is recommended. 14 6 ,14 9 The use of bleach in the environment after initial cleaning procedures is effective for additional elimination of environmental bacteria, and it has been sho wn to be the most effective product in eliminating detectable Salmonella organisms from hospital surfaces. 1 3 7 Several guidelines for the use of different disinfectants and disinfection techniques for materials, stalls and horse facilities have been publish ed. 137 ,14 6 ,1 50 ,15 1 Traffic control measures have been used to prevent and control the spread of Salmonella organisms. Elements of traffic control include the designation of individuals to deal with infected/isolated animals only, restricting the number of hospital personnel or attendants entering isolation stalls, the cleaning of healthy horse stalls before cleaning stalls of sick animals, and control of excessive and unnecessary movement between and through barns. 2 Traffic should always flow from cleaner t o less clean areas. Other management practices recommended for controlling microbial contamination include: use of separate equipment (thermometers, nasogastric tubes, twitches) and cleaning tools (grooming tools, manure carts, forks, brooms, and shovels) for suspicious or confirmed Salmonella positive animals and their stalls. 1 5 2 Animals should not be moved from one stall to another without due consideration for infection control protocols. 14 6 Guidelines for surveillance and infection control Guidelines fo r surveillance and infection control programs that are tailored to the needs and limitations of veterinary hospitals have been published. 92 ,132,13 6 ,14 6 ,15 3 15 5 The following is a summary of surveillance and infection control procedures at the UF LAH.

PAGE 53

53 The o bjective is to reduce the risk of an outbreak of nosocomial Salmonella infections in horses and food animals. The UF LAH has an infection control committee and an infection control officer (ICO). The committee consists of 10 members including clinicians, a microbiologist, an epidemiologist, hospital personnel (veterinary technicians), and the ICO. The committee meets quarterly and its responsibilities include formulation, implementation, and evaluation of hospital surveillance and infection control protocol s and policy. The responsibilities of the ICO include: overseeing sample collection from hospitalized horses and the hospital environment; microbiologic procedures; collection, analysis and reporting of epidemiological data; and the implementation of all i nfection control protocols. In addition, the ICO periodically conducts training and orientations on surveillance and infection control practices to new hospital personnel, faculty, residents, interns, and DVM students. The ICO is supervised by a senior hos pital epidemiologist, who also serves as chair of the infection control committee. All equine patients that present with or later develop signs of gastrointestinal (GI) disease are targeted for early detection of shedding of Salmonella spp in feces at the time of admission and during hospitalization. A fecal sample (or swab) is collected from the rectum of each horse within 12 hours after admission and submitted for bacteriological culture; thereafter, additional samples are collected from the stall floor (or rectum if possible) each morning prior to stall cleaning, every Monday and Thursday during hospitalization. In the past, horses were sampled and tested at admission and every 48 hours (i.e. Monday, Wednesday, and Friday) during hospitalization but bec ause of funding limitations, a maximum of two fecal samples per animal per week

PAGE 54

54 are now collected. Fecal samples are submitted to the microbiology laboratory (located within the hospital complex) for bacteriological culture of Salmonella spp. On average a negative result is reported within 2 3 days after the sample is submitted. In contrast, a positive result is reported within 4 5 days due to additional required testing procedures (e.g., subculturing, serogrouping, and antimicrobial susceptibility testing) In addition to sampling hospitalized horses, routine environmental sampling is carried out monthly or bi monthly to evaluate cleaning and disinfection procedures, or more frequently during periods of high Salmonella activity in the hospital. During each event of routine environmental sampling, 25 of 100 hospital sites are targeted for sampling, and bacteriological culture is used to recover Salmonella from the environmental samples. The selection of sites is not random; instead, sites considered to be at high risk of contamination are sampled. This sample size is considered sufficient to detect (with 95% confidence) one or more sites with evidence of Salmonella contamination if the prevalence of contamination among the 100 sites is 10% or higher, which can be expected during an outbreak situation. 92 Furthermore, stalls used by horses that were Salmonella positive are cultured, and the stall must test negative before other patients are placed in such stalls (unless there is no other stall available, then tha t stall is used after proper cleaning and disinfection procedures have been completed). Also, a new round of environmental samples is collected (eg, n = 25) whenever there is evidence that a nosocomial Salmonella infection has occurred in the hospital, or when a Salmonella positive environmental sample is reported following routine environmental sampling; the objective is to assess cleaning and disinfection

PAGE 55

55 procedures and the magnitude of potential residual environmental contamination of hospital facilities Horses known to be shedding Salmonella spp or that exhibit signs of diarrhea or fever and leucopenia are housed in an isolation unit. Horses that present with signs of GI disease such as colic and do not exhibit signs of diarrhea, or fever and leucopenia and are not known to have tested positive for Salmonella spp are housed in one barn assigned to patients with gastrointestinal disease or the intensive care unit. Horses that present for other complaints such as orthopedic, reproductive, or ophthalmolog ic problems are housed in another barn. Horses that exhibit signs of diarrhea or fever and leucopenia are sometimes housed in the GI barn under ward or stall isolation (i.e., isolated in the individual stall) if the isolation unit is full. To notify hospit al personnel that a stall in the colic barn is under ward isolation, a rope barricade and isolation sign is placed around the stall entrance. Isolation procedures include the use of gloves, plastic boots, gowns, footbaths by all hospital personnel when att ending to patients housed under isolation. Gloves, plastic boots, gowns, and footbaths are placed outside the patient stalls and can be readily accessed. Foot mats containing a quaternary ammonium compound are placed at all points of entry and exit in all barns, the isolation unit, and different hallways within the LAH. Hand hygiene is strongly encouraged including washing of hands and using alcohol based hand disinfectants. Sinks and soap dispensers, and alcohol based hand disinfectant solutions are availa ble at key locations within the hospital. Every time there is evidence of nosocomial Salmonella infection or potential nosocomial Salmonella infection, enhanced infection control measures are instituted

PAGE 56

56 immediately for all hospitalized patients. The enhanc ed infection control measures include mandatory use of gloves, plastic boots, and footbaths by all hospital personnel when attending to every patient in the hospital. In addition, there is restricted movement of patients, for example, patients are not allo wed to get out of stalls to graze or exercise. Rigorous cleaning and disinfection of the hospital (stalls, equipment) is conducted and thereafter, environmental sampling is conducted immediately to assess the magnitude of potential contamination of hospita l facilities and equipment. Enhanced infection control measures are suspended after laboratory and epidemiologic data reveal no further evidence of nosocomial Salmonella infection or hospital environment contamination. Communication is an integral part of a hospital surveillance and infection control program. At the UF LAH, information on the hospital infection control status is relayed quickly and clearly to all stakeholders including clinicians, clients or animal owners, referral veterinarians (RDVM), ho spital personnel, DVM students, the infection control committee, hospital board, and media. Figure 2 2 shows communication channels in place at the UF LAH, where the ICO plays a pivotal role in communication. First, every time an in patient tests positive for Salmonella spp, the laboratory immediately notifies both the attending clinician and the ICO. The attending clinician then informs the client or animal owner and the referral veterinarian about the infection status of the in patient. Clients are educat ed on the potential risk to humans and other animals on the farm, and how to manage that patient after hospital discharge. At the time of discharge, clients are given a Salmonella fact sheet by the attending clinician. The fact sheet includes a brief descr iption of the surveillance program at the hospital, symptoms of clinical

PAGE 57

57 salmonellosis, and measures that should be implemented to reduce the risk of infection of humans and other animals at the farm. If Salmonella positive laboratory results are reported after the patient has been discharged, then the clinician relays the information to clients through a telephone call. In addition, all new hospital clients are given a brochure that describes the hospital surveillance and infection control program. The aim is to inform clients about the surveillance and infection control procedures that are implemented at the hospital to optimize patient care. Second, the ICO makes an assessment to determine if the source of infection is an in patient with a community acqui red infection or a potential nosocomial infection. The ICO establishes communication with the hospital epidemiologist to confirm the source of infection and to define infection control measures that are considered most appropriate. If the patient is classi fied as a primary case, then the ICO communicates and coordinates with hospital personnel to ensure that appropriate infection control measures are implemented for that patient such as isolation and barrier nursing precautions. However, if the patient is c lassified as a potential nosocomial case, then enhanced hospital infection control measures are instituted as previously described. Before implementing enhanced measures, the ICO has to seek approval from at least two members of the infection control commi ttee. Next, the ICO immediately notifies (by email) all clinicians, veterinary technicians, DVM students, and the barn crew about the new infection control status in the hospital and the enhanced measures that must be observed. Third, the ICO is responsibl e for scheduling hospital infection control committee meetings every three months (or more frequently). The main objectives of these

PAGE 58

58 meetings are to provide committee members with a summary report of hospital surveillance and infection control activities f or the past three months and to review and update existing standard operating procedures, as well as the performance of key surveillance parameters (ie frequency of nosocomial cases, compliance in sample collection from horses at admission, rapid reporti ng of laboratory results, efficacy of cleaning and disinfection). Draft meeting minutes are prepared and submitted to all committee members by the infection control committee chair within three days for review and approval. Ten days after the meeting, the minutes are submitted to the hospital board by the committee chair for approval. Finally, when necessary, communication with the media is performed with the assistance of the UF College of Veterinary Medicine: Office of Public Relations. Referring veterina rians (RDVMs) and clients/animal owners play an important role i n a hospital surveillance and infection control program Veterinarians participate in the program by referring clients to a tertiary care hospital because of its specialized expertise and need for specialized care. In turn, clients bring high risk patients such as those with signs of gastrointestinal disease that may be targeted in a hospital surveillance and infection control program. Information about a hospital surveillance and infection con trol program and its importance to RDVMs and clients can be delivered through various channels. At the UF LAH, educational efforts include provision of an infection control brochure to all new clients to educate them about the infection control measures th at are instituted at the hospital to optimize patient care. In addition, every time a patient tests positive for Salmonella, the attending clinician immediately updates the affected clients and RDVMs about the patient Salmonella shedding status and

PAGE 59

59 infecti on control measures that were implemented in the hospital for that patient. At the time of discharge, clients with Salmonella positive horses are given a Salmonella fact sheet by the attending clinician. The purpose of the fact sheet is to educate the clie nt about the potential risks of having a Salmonella positive horse on a farm and measures that should be implemented at the farm level to reduce the risk of exposure to humans and other farm animals. An important element of an animal disease surveillance s ystem is evaluation and feedback. Evaluation of the performance of a hospital surveillance program may involve assessing the frequency of nosocomial infections within a given period of time, frequency of environmental contamination with selected pathogens, compliance level by hospital clinicians and personnel, and level of awareness about the program among RDVMs and clients who refer or bring patients to the hospital. To operate an efficient program, it i s important to solicit feedback from all individuals involved in the program including clinicians, hospital personnel, RDVMs and clients. Considering the role played by RDVMs and clients, it is important to know the level of awareness and perceptions about the relevance of a hospital surveillance and infecti on program among this group. This information can be used to guide decision making on hospital policy issues related to surveillance and infection control and streamlin e costs and efficiency of services provided to clients. Epidemiologic Research Identif ication of risk factors associated with nosocomial Salmonella infection in hospitalized horses is important, so that effective preventative and control measures can be instituted to reduce the risk of disease transmission and potential outbreaks. Initial s tudies conducted in the 1980s and 1990s in California provided an initial epidemiologic framework for investigation of risk factors associated with nosocomial

PAGE 60

60 Salmonella infection in hospitalized horses. A first study investigated an outbreak of nosocomial salmonellosis due to S. Saintpaul in a veterinary hospital. 1 In this study, a presenting complaint of colic, nasogastric intubation, and treatment with antibiotics were associated with isolation of S. Saintpaul. In a second study conducted at the same hos pital, a presenting complaint of colic, nasogastric intubation, and treatment with antibiotics were again associated with isolation of Salmonella. 15 6 Finally, in a third study, diagnosis of large colon impaction, withholding feed, number of days fed bran m ash, treatment with potassium penicillin G, and increase in mean daily ambient temperature were identified as risk factors for nosocomial Salmonella infection. 1 31 A recent epidemiologic study identified abdominal surgery as a risk factor for nosocomial Sa lmonella infection in horses. 1 32 There are several explanations for the observed association between abdominal surgery and nosocomial Salmonella infection in horses. In horses that undergo abdominal surgery, the large intestine may be evacuated and lavage d, feed may be withheld or changed, antimicrobial drugs are usually administered, and various degrees of ileus may develop. 1 31 ,1 5 7 1 6 1 These events can cause stress in surgical patients and alter the gastrointestinal physiology and microflora. In humans an d mice, abdominal surgery has been associated with severe alterations of host defense mechanisms. 1 6 2 16 6 These alterations suppress the innate immune system during the perioperative period and cause substantial impairment of cell mediated immunity. In view of the effects of abdominal surgery on the immune system in humans and mice, it is possible that surgical stress similarly suppresses the innate and adaptive immune systems of equine surgical patients. As a result, abdominal surgery may increase their sus ceptibility to nosocomial Salmonella infections. Studies

PAGE 61

61 are necessary to confirm if abdominal surgery is a predisposing factor for nosocomial Salmonella infection in hospitalized horses. While a high caseload has been suspected as a predisposing factor fo r nosocomial Salmonella infection in hospitalized horses, previous epidemiologic studies 1,131 ,1 32 ,15 6 failed to identify high caseload as a risk factor. A study in Florida 1 32 explained that the lack of an association between a high caseload and nosocomial Salmonella infections could be attributed to the fact that the number of horses shedding Salmonella at the time of admission during periods of high or low caseload was not different. Another explanation was that the hospital surveillance and infection cont rol program was well established and the degree of personnel compliance was acceptable. It is possible that high caseload may have an effect on risk of nosocomial infection in hospitalized horses when the number of shedders in the hospital is high and infe ction control standards are suboptimal. In veterinary hospitals, horses that present with diarrhea at admission or that develop diarrhea during hospitalization are placed in isolation to reduce the risk of Salmonella transmission to other inpatients. Three p ublished studies have showed, however, that the frequency of hospitalized horses that were classified as positive to Salmonella shedding and did not have diarrhea at admission or during hospitalization can be high. For example, in one study, among 33 hor ses from which Salmonella S aintpaul was isolated, 27 (88%) did not have diarrhea. 1 In another study, among 131 horses from which Salmonella enterica was isolated, 125 (95%) did not have diarrhea. 3 Finally, in a third study, among 233 horses from which Salm onella enterica was isolated, 197 (85%) did not develop diarrhea. 16 7 S urveillance data collected at the

PAGE 62

62 University of Florida Large Animal Hospital (UF LAH) over the last four years (2007 2010) revealed that 52 of 67 adult equine inpatients with signs of G I disease that tested positive for Salmonella did not have diarrhea at admission or during hospitalization. Horses that test positive for Salmonella and do not have diarrhea at admission or during hospitalization present a threat because most horses are de tected after discharge and there is no opportunity to implement infection control measures during hospitalization. This subpopulation of horses can also present a threat to other horses and humans in the community when the equine patient returns home. Expo sure factors associated with Salmonella shedding in horses with or without diarrhea have not been investigated. Information on exposure factors associated with Salmonella shedding in horses without diarrhea is important because it can help improve current hospital surveillance programs for early detection of Salmonella shedding in GI horses without diarrhea and allow timely implementation of infection control measures. Discussion In previous outbreaks where hospitals had to close, the lack of a case definit ion for nosocomial Salmonella infections, day to day monitoring of nosocomial cases, routine environmental sampling, infection control protocols, and guidelines for hospital closure due to an increased number of nosocomial Salmonella infections, all contri buted to the onset of outbreaks of Salmonella infections in hospitalized horses. A question that is dreaded and yet warrants discussion by directors of hospital infection control programs and hospital administrators is when to temporarily close a hospital due to an increased number of nosocomial infections in horses. Currently, there are no published reports with standardized criteria or guidelines to decide when a hospital should be temporarily closed to stop a nosocomial outbreak of Salmonella infections in hospitalized horses. A

PAGE 63

63 review of previous outbreak reports revealed several parameters that can be taken into consideration (Table 2 2 ) The consequences of previous nosocomial outbreaks of Salmonella infections in horses clearly justify the need to for mulate standards for hospital surveillance and infection control measures, including specific guidelines for when to close a hospital operation to prevent unnecessary morbidity and mortality in hospitalized horses. Three key parameters that can be used by hospital administrators when making a decision to temporarily close a veterinary hospital include: (i) weekly number of nosocomial Salmonella cases with or without clinical disease; (ii) case fatalities; and (iii) zoonotic disease. Exposure factors associa ted with Salmonella shedding in horses with or without diarrhea have not been investigated. Previous epidemiological studies identified several factors associated with Salmonella shedding in horses including colic, 1,131,15 6 ,16 8 transportation, 1 60 ,16 9 clini cal procedures such as change in diet, 16 7 antimicrobial therapy, 152 ,1 60 ,1 70 abdominal surgery, 14 5 ,15 2 surgically treated small colon impactions, 17 1 clinical hospitalization, 16 9 ,16 7 fever during hospitalization, 16 7 hospitalization and abnormal results of nasogastric intubation. 16 9 In addition, f our studies identifi ed exposure factors associated with nosocomial Salmonella infection in hospitalized horses including a presenting complaint of colic, 1,156 nasogastric intubation, 1,156 treatment with antibiotics, 1,156 diagnosis of large colon impaction, 131 withholding feed 131 number of days fed bran mash, 131 treatment with potassium penicillin G, 131 increase in mean daily ambient temperature, 131 and abdominal surgery. 132 However, in all those previous studies, the analyses did not separate horses

PAGE 64

64 with and without diarrhea T herefore the factors identified in those studies apply to a combined population of horses with and without diarrhea. I nformation on exposure factors associated with Salmonella shedding in horses without diarrhea is important because it can help improve c urrent hospital surveillance programs for early detection of Salmonella shedding and allow timely implementation of infection control measures. Although time to obtain Salmonella test results by bacteriologic culture is a limitation, bacteriologic culture is a good diagnostic tool to use in hospitals where compliance with surveillance and infection control protocols is good. Clinicians often argue that bacteriologic test results have limited value for patient care because laboratory results are often report ed days after the inpatient has been discharged and other patients may have been exposed by the time laboratory results are reported. While this argument is valid, hospital surveillance and infection control protocols are designed to manage the infection c ontrol status of a hospital rather than that of individual inpatients. The proposed use of PCR protocols as a surveillance tool for early detection of Salmonella fecal shedding in horses is a good idea, provided the frequency of false positives is low, ac ceptable, and does not significantly affect the cost of hospital fees. The study by Ward et al 8 justifies the need to further investigate the reasons why horses without clinical signs of salmonellosis that are classified as negative (i.e., after five cons ecutive fecal samples using bacteriological culture) test positive by PCR. In addition, for PCR to be accepted as a surveillance tool, th e issue of non objective sampling methods used in previous PCR studies 10,13 needs to be addressed in future studies. D ifferent sampling methods can affect the diagnostic performance assessment

PAGE 65

65 of diagnostics tests. Wh ile assessing specificity important that a gold standard definition for true negatives that considers horses without clinical signs of salmonellosis th at are sampled multiple times is used. Bacteriological culture of multiple fecal samples from horses has been shown to yield a greater proportion of Salmonella positive results compared to single fecal samples. 1 40 In an effort to standardize hospital surv eillance and infection control programs and to institute a measure of accountability, the reporting of nosocomial Salmonella infections in hospitalized horses (and other species) should be considered as one parameter of excellence in hospital veterinary ca re. Reporting of nosocomial Salmonella infections may encourage improvement in the quality of patient care, and the overall efficiency of a hospital surveillance and infection control program. In light of the recent epidemiologic finding 1 32 that abdominal surgery was identified as a risk factor for nosocomial Salmonella infection in hospitalized horses, there is need to further investigate this finding to confirm if surgical stress suppresses the innate and adaptive immune systems of equine surgical patient s, thereby increasing their susceptibility to nosocomial Salmonella infections. Results from additional studies will help determine if abdominal surgery is a predisposing factor for nosocomial Salmonella infections in hospitalized horses, and justify the n eed to enhance infection control measures (eg, isolation and use of gloves, gowns, plastic boots, and footbaths) to protect equine inpatients that undergo abdominal surgery. Referring veterinarians and clients play an important role in a hospital surveill ance and infection control program. Considering the role played by RDVMs important to identify and characterize their level of awareness and perceptions about

PAGE 66

66 the relevance of a hospital surveillance and infection control program. No stud ies have been conducted to assess the level of awareness and relevance of a hospital surveillance and infection control program among referring veterinarians and clients. This information can be used to justify the need for enhanced education efforts among referring veterinarians and clients and guide decision making on policy issues related to hospital surveillance and infection control The following knowledge gaps were identified after a critical review of the literature o n epidemiology and infection c ontrol of nosocomial Salmonella infections in hospitalized horses : 1. T here is no standard case definition for nosocomial Salmonella infections that can be used as a guideline when implementing surveillance and infection control procedures. 2. T here are no publ ished reports with standardized criteria or guidelines to decide when a hospital should be temporarily closed to stop a nosocomial outbreak of Salmonella infections in hospitalized horses. 3. E xposure factors associated with Salmonella shedding in horses wi th or without diarrhea have not been investigated. 4. F or PCR to be accepted as a surveillance tool, I t is important to assess the diagnostic performance of a PCR assay using objective sampling methods; with horses defined as true negatives used to estimate specificity that are sampled multiple times In addition, the PCR assay should have a low frequency of PCR false positive results that is considered acceptable. 5. F urther studies to confirm if surgical stress suppresses the innate and adaptive immune systems of equine surgical patients need to be conducted 6. T here is need to assess the level of awareness and relevance of a hospital surveillance and infection control program among referring veterinarians and clients. Three of the knowledge gaps identified in this literature review were addressed in this research work. First, an investigation of the exposure factors associated with Salmonella shedding in hospitalized horses with or without diarrhea was conducted.

PAGE 67

67 Second, an investigation was performed to assess the diagnostic performance of a real time PCR assay for detection of Salmonella spp in fecal specimens of hospitalized horses, compared to bacteriological culture. Finally, a survey was performed to assess the level of awareness and relevance of a hospita l surveillance and infection control program among referring veterinarians and clients.

PAGE 68

68 Table 2 1. Reported sensitivity and specificity estimates of real time PCR protocols for detection of Salmonella spp in equine fecal samples Primer probe Sensitivit y % (# samples tested) Specificity % (# samples tested) Reference spaQ gene 100.0 (80) 97.3 (150) Kurowski et al. 2002 12 invE A gene 80.0 (45) 98.6 (729) Gentry Weeks et al. 2002 9 hisJ gene 93.0 (15) 85.5 (291) Gentry Weeks et al. 2002 9 invA gene 100.0 (28) 98.3 (345) Bohaychuck et al. 2007 13 invA gene 100.0 (6) 98.2 (895) Pusterla et al. 2009 10 Table 2 2. Large animal hospitals that have closed because of an outbreak of nosocomial Salmonella infections and surveillance parameters of interest: Hosp ital Duration of hospital closure Parameters that might have influenced the decision to close the hospital Reference 1 3 months Increased number of nosocomial cases in horses: 33 cases diagnosed with S. Saintpaul. 20 of 33 cases showed clinical signs of salmonellosis. 2 of the 33 cases died. Hird et al 1984 1 2 3 months 34 of 137 food animal and equine inpatients tested positive for the outbreak strain: S. Infantis. Tillotson et al. 1997 2 3 1 month 18 of 138 inpatients tested positive for the outbreak s train: S. Typhimurium. 8 of 18 affected horses were euthanized or died. 1 zoonotic infection reported. Schott et al. 2001 3 4 3 months 28 horses tested positive for the outbreak strain S. Typhimurium. 14 horses infected with S. Typhimurium died or were eut hanized. Ward et al. 2005 4

PAGE 69

69 Figure 2 1. A diagram showing two potential sources of nosocomial infection: an inpatient index case or environmental contamination.

PAGE 70

70 Figure 2 2. Surveillance and infection control communication channels at the University of Florida, Large Animal Hospital.

PAGE 71

71 CHAPTER 3 EXPOSURE FACTORS ASS OCIATED WITH SALMONE LLA SHEDDING AMONG HOSPITALIZED HORSES WITH OR WITHOUT DIAR RHEA Despite implementation of biosecurity standard operating procedures, veterinary hospitals have experienc ed outbreaks of salmonellosis causing high mortality in equine inpatients and substantial financial losses. 1 6 As a result, epidemiologic studies have been conducted to identify exposure factors associated with Salmonella shedding in hospitalized horses an d to help formulate and implement improved biosecurity standard operating procedures. Several studies have identified hospitalized horses with colic to be at high risk of shedding Salmonella 1, 131,15 6 ,16 8 Other factors associated with Salmonella shedding i nclude transportation, 1 60 ,16 9 clinical procedures such as change in diet, 16 7 antimicrobial therapy, 15 2 ,1 60 ,1 70 abdominal surgery, 1 4 5 ,15 2 surgically treated small colon impactions, 1 7 1 after hospitalization, 1 6 7 ,1 6 9 fever during hospitalization, 1 6 7 after hospitalization and abnormal results of nasogastric intubation. 1 6 9 In addition, f our studies identified presenting complaint of colic, 1,15 6 nasogastric intubation, 1,1 5 6 treatment with antibiotics, 1,15 6 diagnosis of large colon impaction, 131 withholding feed, 131 number of days fed bran mash, 131 treatment with potassium penicillin G, 131 increase in mean daily ambient temperature, 131 and abdominal surgery as exposure fact ors associated with nosocomial Salmonella infection in hospitalized horses. 132 In veterinary hospitals, horses that present with diarrhea at admission or that develop diarrhea during hospitalization are placed in isolation to reduce the risk of Salmonella transmission to other inpatients. Three p ublished studies have showed, however, that the frequency of hospitalized horses that were classified as positive to Salmonella shedding and did not have diarrhea at admission or during hospitalization

PAGE 72

72 can be high. For example, in one study, among 33 horses from which Salmonella S aintpaul was isolated, 27 (88%) did not have diarrhea 1 In another study, among 131 horses from which Salmonella enterica w ere isolated, 125 (95%) did not have diarrhea 3 Finally, in a third study, among 233 horses from which Salmonella enterica was isolated, 197 (85%) did not develop diarrhea 1 6 7 S urveillance data collected at the University of Florida Large Animal Hospital (UF LAH) over the last four years (2007 2010) revealed that 52 of 6 7 adult equine inpatients with signs of gastrointestinal (GI) disease (colic or diarrhea) that tested positive for Salmonella did not have diarrhea at admission or during hospitalization. On average, duration of hospitalization for equine inpatients with s igns of GI disease at the UF LAH is 5 days and bacteriologic culture procedures for detection of Salmonella require 4 5 days before a positive result is reported. Using existing hospital surveillance protocols, horses that test positive for Salmonella and do not have diarrhea at admission or during hospitalization present a threat because most horses are detected after discharge and there is no opportunity to implement infection control measures during hospitalization. This subpopulation of horses can also present a threat to other horses and humans in the community when the equine patient returns home. To our knowledge, exposure factors associated with Salmonella shedding in horses with or without diarrhea have not been investigated. This information is im portant as it can further improve current hospital surveillance programs for early detection of Salmonella shedding and permit rapid implementation of infection control measures. The objective of this study was to examine the relationships between clinical signs, hematological and plasma chemistry parameters, as well as clinical procedures (before

PAGE 73

73 admission, at admission, and during hospitalization) and Salmonella shedding in adult horses with signs of GI disease that tested positive for Salmonella and had or did not have diarrhea, compared to adult horses with signs of GI disease that tested negative for Salmonella and did not have diarrhea. Materials and Methods Study P opulation All equine patients admitted to the UF LAH between January 1, 2007, and Decemb er 31, 2010, with signs of GI disease ( colic; diarrhea; diarrhea and fever; diarrhea and anorexia; anorexia and fever; anorexia ) that were tested for Salmonella enterica shedding in fecal specimens were initially considered for inclusion in the study. Foal s (< 1 year old) were excluded. The rationale for excluding foals was that exposure factors for diarrhea in foals can differ from those for adult horses. Furthermore, UF LAH surveillance data for the study period revealed that nearly all foals admitted wit h signs of GI disease were foals affected with diarrhea upon admission, limiting our ability to study an adequate number of foals without diarrhea. Horses classified as nosocomial infections (n=6; serotypes isolated were Mbandaka, Miami, Litchfield, Meleag ridis, Newport, and a non identified serotype) were excluded because our interest was to extrapolate study results to equine in patients with evidence of community acquired Salmonella infections only. The criteria used to define nosocomial Salmonella infec tions in this study have been reported in detail elsewhere. Hospital Surveillance a nd Infection Control Procedures During the study period, all equine patients admitted to the UF LAH with signs of GI disease (colic or diarrhea or anorexia or fever ) were t argeted for early detection of fecal shedding of Salmonella enterica at the time of admission and during

PAGE 74

74 hospitalization. Details of the surveillance and infection control procedures at the UF LAH are described elsewehere. 6,132 Briefly, a fecal sample or s wab (if fecal sample is not available) i s collected per rectum from each horse within 12 hours after admission and submitted for bacterial culture. Thereafter, additional fecal samples a re collected from the stall floor each morning prior to cleaning at 48 hour intervals until the patient i s discharged from the hospital. Fecal samples collected outside regular business hours a re refrigerated at 4 C prior to laboratory submission. For some horses, additional fecal samples a re collected at the discretion of t he attending clinician (eg, every 12 to 24 hours). Any equine patient that test s positive for Salmonella enterica or develop s diarrhea, or fever and leucopenia i s placed in an isolation barn. Isolation procedures include use of barrier nursing precautions (such as gloves, plastic boots, gowns, and footbaths) for personnel attending to patients. In addition to sampling hospitalized horses, routine environmental sampling is carried out monthly or bi monthly to evaluate cleaning and disinfection procedures, or more frequently during periods of high Salmonella enterica activity in the hospital. During each event of routine environmental sampling, 25 hospital sites are targeted for sampling, and bacteriological culture is used to recover Salmonella enterica from the environmental samples. A hospital infection control officer i s responsible for overseeing collection of fecal samples, microbiologic procedures, and collection and analysis of epidemiologic data. Study Design A case control study approach was used to examine the relationships between clinical signs, hematological and plasma chemistry parameters, and clinical procedures before admission, at admission (day 1), and during hospitalization (days 2 and 3) and Salmonella shedding in GI horses with or without diarrhea. Exposure factors during

PAGE 75

75 hospitalization were measured on hospital days 2 and 3 only because duration of hospitalization was different among study horses and testing procedures (eg, hematological and plasma chemistry parameters) were not required by the attending clinician in most horses after day 3. Selection of cases Two sets of case horses were used in this study: case horses that presented with signs of GI disease with or without diarrhea and tested positive for Salmonella enterica Cases with diarrhea (n = 19) were further divided into 2 groups: (i) horses that tested positive for Salmonella enterica on fecal samples collect = 14), and (ii) horses that tested negative on the first two fecal samples but positive on of horses tested negative on at least 2 conse after admission. Diarrhea was defined as 3 consecutive bowel movements in which feces do not sit on top of the bedding. Cases without diarrhea (n = 52) were also divided into 2 groups: (i) horses that tested positive for Salmonella enterica admission (n = 25) and (ii) horses that tested negative on the first two fecal samples but Six case horses without diarrhea did not meet these inclusion criteria and were excluded because they tested negative on only The reason for creati ng the two subgroups of case horses with or without diarrhea that tested positive for Salmonella enterica

PAGE 76

76 investigated exposure factors and Salmonella shedding at two defined exposure periods of interest: a) before admission and at admission (day 1) and b) on days 2 and 3 after admission. Selection of controls Control horses were defined as horses that tested negative for Salmonella enter ica on 3 or more consecutive fecal samples and did not have diarrhea before or during the entire period of hospitalization. Control horses were randomly selected based on year of admission. Selection of number of control horses was based on the total numbe r of cases horses without diarrhea (n = 52) diagnosed during 2007 2010 (ie, 10 cases in 2007, 11 in 2008, 18 in 2009, and 13 in 2010). Up to 2 control horses per case were selected in 2007 and 2008 and one control horse per case horse in 2009 and 2010 A l ower caseload in 2009 and 2010 limited our ability to select two controls for each case horse in those two years. Thus, the ratio of controls to cases was 1:1. The final enrollment included 73 controls: 20 from year 2007, 22 from 2008, 18 from 2009, and 13 from 2010. The program research randomizer was used to generate random numbers and select control horses. a Microbiologic Procedures f or Detection o f Salmonella Spp Bacterial culture of fecal samples for detection of Salmonella enterica was performed at th e UF Veterinary Clinical Microbiology Laboratory as previously described. 6, 132 Briefly, for selective enrichment, 1 to 2 g of fresh feces or a rectal swab was placed in 10 mL of selenite cystine broth, and incubated at 37 C for 24 hours. The following day, a sample of the selenite cystine broth was subcultured on Hektoen a Research Randomizer, Social Psychology Network, Middletown, Conn. Available at: www.randomizer.org ; a ccessed March 15 th 2011

PAGE 77

77 enteric agar plates. Plates were incubated at 37 C for 24 hours. Non lactose fermenting, hydrogen sulfide producing colonies were selected and isolated. These colonies were then inoculated on urea agar and lysine iron agar slants and incubated at 37 C for 24 hours. Identification of urease negative and hydrogen sulfide producing organisms was established through use of a commercially available identification system. Serogroup of Salmonella isolates was determined by means of agglutination; polyvalent (A through I and Vi) and group specific (A through E) Salmonella O antisera were used. Salmonella isolates were tested for antimicrobial susceptibility using the minim al inhibitory concentration method with commercially prepared plates and the Kirby Bauer disk diffusion method. Serotyping of Salmonella isolates was performed at the USDA National Veterinary Services Laboratories in Ames, Iowa. Data Collection A questionnaire was developed to colle ct epidemiologic data for each study horse during three exposure periods of interest: before admission, at admission (day 1), and during hospitalization (days 2 and 3). Briefly, data collected before admission included host factors (patient age, sex, breed ), clinical findings (diarrhea (yes/no), fever (yes/no), leucopenia (yes/no), and clinical procedures (history of antimicrobial use (yes/no; types used), anti inflammatory use (yes/no; types used), sedative use (yes/no; types used), antispasmodic use (yes/ no; types used), nasogastric intubation (yes/no), laxative use (e.g., mineral oil) (yes/no; types used), stomach protectant use (e.g., omeprazole) (yes/no; types used), intestinal protectant use (e.g., di tri octahedral smectite) (yes/no; types used), IV f luid use (yes/no; types used), and rectal exam findings). Data collected at admission included admission date, season of year at admission

PAGE 78

78 (winter/spring/summer/fall), presenting complaint, clinical findings (diarrhea, fever, leucopenia, heart rate (beats/ min), color of mucous membranes (pink/dark pink/injected/icteric), capillary refill time, plasma chemistry and complete blood count results (normal/low/high), and clinical procedures (nasogastric intubation, rectal exam findings, abdominal ultrasound findi ngs). Data collected during hospitalization included clinical findings (diarrhea, fever, leucopenia, plasma chemistry and complete blood count results), clinical procedures (nasogastric intubation, rectal exam findings, abdominal ultrasound findings, type of feed given, withholding of feed (yes/no; duration), laxative use, stomach protectant use, intestinal protectant use, IV fluid use, general anesthesia (yes/no), abdominal surgery (yes/no), anatomic location of surgical lesion, type of surgical lesion, ty pe of surgical procedure, antimicrobial use, anti inflammatory use, sedative use, antispasmodic use, final diagnosis), discharge date, duration of hospitalization, and number of fecal samples collected. For case horses, information on Salmonella isolates w as collected including serogroup, serotype, fecal sample number that first tested positive, date when first positive fecal sample was reported, and number of hospital days when first test positive sample was collected. Data Analysis The Kruskal Wallis tes t b was used to compare continuous variables (number of fecal samples collected and duration of hospitalization) among case horses with diarrhea, case horses without diarrhea, and control horses. b PROC NPAR1WAY, SAS 9.2, SAS Institute Inc, Cary, North Carolina.

PAGE 79

79 Unconditional logistic regression was used to model the odd s of Salmonella shedding (yes, no) using the LOGISTIC procedure in SAS. c In the univariable analysis, variables with a P value < 0.10 were considered eligible for multivariable analysis. To determine the best fitting model, the variable with the smallest P value in the univariable analysis was entered into the model first. Thereafter, each of the remaining variables was added to the model containing the first variable to test whether its addition significantly improved the fit of the model. The variable wit h the highest likelihood ratio statistic ( X 2 test with one degree of freedom) was selected for addition to the model and the process was then repeated; variables had to have a P 1 0 to be retained in the model. To determine whether the main effect s model was continuous in the logit for the variable of number of samples tested, we examined the coefficients using indicator variables based on quartiles of the distribution. Because of the small sample size used in each model, interaction terms between exposure factors were not tested. In Models 2 and 3, exposure variables retained in the model were examined for confounding by adding each of the variables to the model and assessing the changes in the odds ratios (ie, > 20%) of the remaining variables in the model. The logistic regression model was assessed for goodness of fit using the Hosmer Lemeshow test, while the ability to discriminate between horses that tested positive to Salmonella shedding versus those that did not was determined by using the rec eiver operating characteristic (ROC) curve. In the final model, adjusted OR and 95% confidence intervals were reported. To accomplish the study objective, three models were developed. Model 1 included 14 case horses that tested positive for Salmonella on f c PROC LOGISTIC, SAS 9.2, SAS Institute Inc, Cary, North Carolina.

PAGE 80

80 2 days after admission and had diarrhea. Model 2 included 25 case horses that tested positive for Salmonella have diarrhea. Model 3 included 21 case horses that tested positive for Salmonella on model, the variable for number of samples tested was forced in the model because it can increase the likelihood of detecting horses that t est positive to Salmonella shedding. In each of the three models examined, the same set of control horses was used. A fourth model of case horses that tested positive for Salmonella on fecal samples as not examined because the sample size was too small (n=5). In this study, the number of case horses used was small; therefore, we limited the number of variables offered to each of the final models (based on the number of cases). Use of number of events per variable less than 10 can lead to biased parameter estimates in a logistic regression model. 17 2 Results The median age for case horses with diarrhea (n=14) was 12 years (1 st quartile = 4.5 & 3 rd quartile = 17.25); for case horses without diarrhea (n=4 6) was 9.5 years (6 and 13); and for control horses (n=73) was 8 (6 and 12). Among case horses with diarrhea (n=14), 9 were female, 1 was male, and 4 were geldings. Among case horses without diarrhea (n=46), 19 were female, 2 were male, and 25 were gelding s. Among control horses (n=73), 33 were female, 8 were male, and 32 were geldings. Among case horses with diarrhea (n=14), the most common breeds were Thoroughbred (3), American Paint horse (3), Arabian (3), Quarter Horse (2), and others

PAGE 81

81 (3). Among case horses without diarrhea (n=46), the most common breeds were Thoroughbred (11), Quarter Horse (11), Warmblood (7), American Paint horse (3), Arabian (3), and others (11). Among control horses (n=73), the most common breeds were Quarter Horse (19), Thoroughb red (16), Warmblood (9), American Paint horse (6), Arabian (3), American Miniature (3), Pasofino (3), and others (14). Among case horses with diarrhea (n=14), 7 presented for colic, 3 for diarrhea, 2 for diarrhea and fever, 1 for diarrhea and anorexia, and 1 for anorexia and fever. Among case horses without diarrhea (n=46), 45 presented for colic and 1 for anorexia. Among control horses (n=73), 71 presented for colic and 1 for anorexia and fever, and 1 for anorexia. Among case horses with diarrhea (n=14 ), 11 were medically treated and 3 were surgically treated. Among case horses without diarrhea (n=46), 25 were medically treated and 21 were surgically treated. Among control horses (n=73), 35 were medically treated and 38 were surgically treated. Among 14 case horses with diarrhea, the most commonly isolated Salmonella enterica serotypes were Anatum (n = 2) and Muenchen (n=2) (Table 3 1 ). Among 46 cases without diarrhea, the most commonly isolated serotype was Newport (n=13). In 2007, 12 isolates were rep orted and the most common serotype was Anatum (n=2). In 2008, 17 isolates were reported and the most common serotypes were Anatum (n=3), Newport (n=3), Miami (n=2), and Saintpaul (n=2). In 2009, 19 isolates were reported and the most common serotypes were Newport (n=6), Muenchen (n=3) and Anatum (n=2). In 2010, 12 isolates were reported and the most common serotypes were Newport (n=3) and Muenchen (n=2).

PAGE 82

82 The median number of fecal samples collected was not different between cases with diarrhea (3 samples; 1 st quartile = 3 & 3 rd quartile = 5), cases without diarrhea (3.5 samples; 2.75, 5), and controls (3 samples; 3, 4) ( P = 0.66). The median duration of hospitalization was not different between cases with diarrhea (5 days; 1 st quartile = 4 & 3 rd quartile = 7 .5; range = 2, 10), cases without diarrhea (7 days; 4 and 9; 2, 39), and controls (6 days; 5 and 7; 1, 10) ( P = 0.48). The median number of days of hospitalization when a horse was first identified as positive was 2.5 days (1 st quartile = 1 & 3 rd quartile = 4; range = 1, 6) for cases with diarrhea and 3 days (2 and 6; 1, 12) for cases without diarrhea. The median number of the fecal sample that first tested positive was significantly different between cases with diarrhea (median = 1.5; 1 st quartile = 1 & 3 rd quartile = 3) and cases without diarrhea (3; 2, 3) ( P = 0.01). No horses with or without diarrhea had both fever and leucopenia at admission. Positive Salmonella culture results for 32 of the 46 case horses without diarrhea were completed and reported after discharge. Among case horses with or without diarrhea that tested positive to Salmonella shedding, the median number of days between time of fecal sample collection and reporting of results for case horses was 5 days (1 st quartile = 5, 3 rd quartile = 6; range 4, 14) for positive samples and 2 days for negative samples (2 and 6; range 2, 19). Model 1: Case Horses t hat Tested Positive f or Salmonella a a fter Admission a nd h ad Diarrhea In the univariable analysis, the following variables were as sociated with Salmonella shedding ( P 0.05) (Table 3 2): Before admission: antimicrobial and anti inflammatory use. At admission: high triglycerides (> 45 mg/dL), low plasma sodium (<

PAGE 83

83 134 mEq/L), low plasma protein (< 5.9 g/dL), and high mean corpuscular hemoglobin concentration (> 38.8 g/dL). Multivariable analysis was not performed for model 1 because of the small number of cases. Model 2: Case Horses t hat Tested Positive f or Salmonella a fter Admission a nd d id n ot h ave Diarrhea In the univaria ble analysis, the following variables had values of P 0. 10 (Table 3 3). Before admission: sedative use. At admission: high blood urea nitrogen (> 22 mg/dL), high triglycerides, high plasma protein (> 7 .9 g/dL) high red blood cell count (> 10 M/uL), and neutropenia (< 2.2 K/uL). In the multivariable analysis, high plasma protein concentration at admission (OR = 2.85; 95% CI = 0.91, 8.91; P = 0.07; Table 3 5) and high triglycerides at admission (OR = 3.37; 95% CI = 1.15, 9.87; P = 0.02; Table 3 5) were a ssociated with Salmonella variable year of admission was added to the final model. The model fit the data well as indicated by the Hosmer Lemeshow goodness of fit test (0.73; df = 4; P = 0.94). However, the area under the curve for the ROC was 0.67 indicating that the model has low predictive ability. In this model, the variable number of samples tested was not forced into the final models because the median number of samples among cases and controls was the same (median = 3). Model 3: Case Horses that Tested Positive f or Salmonella a a fter Admission a nd d id n ot h ave Diarrhea, and Tested Negative o n a t l east 2 Fe cal fter Admission In the univariable analysis, the following variables had values of P 0.10 (Table 3 4). At admission: high alkaline ph osphatase (> 228 U/L), high creatine kinase (> 729 U/L), low chloride (< 96 mEq/L), high mean corpuscular hemoglobin (> 18.7 pg), high

PAGE 84

84 corpuscular hemoglobin (> 17.9 pg), elevated heart rate (41 60 beats per min), and season. During hospitalization: antim icrobial use, leucopenia, fever and leucopenia, withholding of feed, stomach protectant use, general anesthesia, abdominal surgery, low plasma protein on hospital day 2, and highly elevated heart rate on day 3 (> 60 beats per min). The variables for mean c orpuscular hemoglobin and corpuscular hemoglobin were correlated; therefore only mean corpuscular hemoglobin was selected for inclusion in the multivariable model. In the multivariable analysis, two final models were examined. In the first model (Model A; Table 3 6), after adjusting for number of samples, the variable for abdominal surgery was retained in the final model for Salmonella admission in horses without diarrhea (OR = 2.79; 95% CI = 0.86, 9.01; P = 0.08). The model fit the data well, although the Hosmer Lemeshow statistic almost reached statistical significance (8.78; df = 4; P = 0.06). The area under t he curve for the ROC was 0.73 indicating that the model has moderate predictive ability. No confounding effect was observed when the variable for year of admission was added to the final model. In the second model (Model B; Table 3 6), after adjusting for number of samples, season (summer) (OR = 4.92; 95% CI = 1.46, 16.57; P = 0.01) was associated with Salmonella sh by the Hosmer Lemeshow goodness of fit test (4.97; df = 4; P = 0.28). The area under the curve for the ROC was 0.77 indicating that the model has moderate predictive ability. No conf ounding effect was observed when the variable for year of admission was added to the final model.

PAGE 85

85 Discussion The present study examined the relationships between clinical signs, hematological and plasma chemistry parameters, and clinical procedures, and S almonella shedding in GI horses with or without diarrhea. Following a univariable analysis, a mong horses that tested positive for Salmonella had diarrhea, use of antimicrobials and anti inflammatories before admission, high plasma triglycerides, low plasma sodium, low plasma protein, and high mean corpuscular hemoglobin concentration at admission were associated wi th Salmonella shedding. Following a multivariable analysis, a mong horses that tested positive for Salmonella concentration and high plasma triglycerides at admission were associated with Sa lmonella shedding. Among horses without diarrhea that tested positive for Salmonella Salmonella shedding. Our study had several limitations. First, the sample size used in this st udy was small and it limited the number of exposure factors that could be examined in the epidemiologic analyses, and it also affected the precision of the estimated associations between exposure factors measured and Salmonella shedding. Second, bacteriolo gic culture procedures used in the present study differed from those used in other studies. 1 6 Methods used to culture Salmonella are highly variable among laboratories, 132 ,1 3 8 with differences such as weight of fecal sample which affect the sensitivity of bacteriologic culture. 13 9 Third, bacteriological culture of multiple fecal samples from horses has been shown to yield a greater proportion of Salmonella positive results compared to single fecal samples. 1 40 This is an inherent limitation in observational studies that are based on

PAGE 86

86 hospital surveillance data. For example, at the UF LAH, 132,15 2 equine inpatients on average remain hospitalized for 5 days and a total of 5 or more fecal samples are rarely collected for detection of Salmonella shedding. Fourth, assessment of e xposure factors during hospitalization was limited to the first 3 days because duration of hospitalization was different among study horses and frequency of testing procedures (eg, hematological and plasma chemistry parameters) was not consi stent in study horses after day 3. Fifth, in this study, pre admission data were collected from history provided by clients, most of these data were missing and the accuracy of this information was not assessed. Thus, it is difficult to determine the magni tude of pre admission exposure misclassification potentially present in this study. Sixth, the study population was limited to adult equine patients with GI disease; therefore our study results cannot be generalized to the entire equine population at the U F LAH. Finally, there are differences between the UF LAH and other veterinary hospitals (ie, climate and housing conditions, surveillance and infection control measures, pathogen infection control pressure) which make the extrapolation of our study finding s to other hospitals difficult. In the present study, among horses without diarrhea, Salmonella enterica serotype Newport was the most commonly isolated serotype. The Newport isolates were susceptible to all antimicrobials tested. Salmonella Newport has be en previously reported in equine inpatients affected with colic but without diarrhea at the UF LAH. 1 32,15 2 In contrast, in two previous outbreak investigations of nosocomial salmonellosis in horses, cattle, and alpacas, multi drug resistant strains of Sal monella Newport were identified as the causative strain at other hospitals. 5, 92 In one study, 17 of 54 horses infected with the outbreak strain died during hospitalization (but it is was not

PAGE 87

87 clear how many horses were affected with diarrhea). 5 In another study, 1 of 2 horses infected with the outbreak strain developed enterocolitis, diarrhea, septicemia and subsequently died. 92 A comparison of this and previous studies 5, 92 shows that infections with drug resistant S Newport strains are associated with diar rhea and mortality in hospitalized horses, compared to susceptible S Newport strains. Several studies in humans have demonstrated that infections with drug resistant non typhoid Salmonella serotypes are associated with excess morbidity and mortality. 97,99 In the present study, among 46 case horses without diarrhea, no horses had fever and leucopenia at admission. This finding is relevant at the UF LAH because (following hospital infection control procedures) horses that exhibit a combination of fever and l eucopenia are housed in an isolation unit. 132 ,1 5 2 This hospital policy is based on the fact that fever and leucopenia are clinical signs associated with Salmonella infection in horses. 17 3 17 6 While this policy is logical, this study shows that its relevan ce among equine inpatients without diarrhea is limited. In this study, 32 of 46 case horses without diarrhea were classified as Salmonella shedders after discharge from the hospital. At the UF LAH, clients and referring veterinarians with horses that are r eported as Salmonella culture positive after discharge are informed about the Salmonella shedding status by the attending clinician by telephone. During the same time, they are educated about the risks of having a Salmonella positive horse on the farm to h umans and other animals, and are advised on how to manage the patient at the farm (including isolation measures). The fact that many Salmonella shedders without diarrhea were detected after discharge and were not placed in isolation during hospitalization justifies the need

PAGE 88

88 to re evaluate the risk based surveillance approach used at the UF LAH for detection of Salmonella shedders in equine inpatients. Model 1: Case horses that tested positive for Salmonella admission and had diarrhea I n the univariable analysis, among horses that tested positive for Salmonella or anti inflammatories, as well as high plasma triglycerides, low plasma sodium, low plasma protein, or high mean corpuscular hemoglobin concentration at admission were associated with Salmonella shedding. The small number of cases in this analysis limited our ability to adequately examine the association between these exposure factors and Salmonella shedding using mu ltivariable logistic regression Another limitation in this analysis is that exposure factors associated with Salmonella shedding in horses with diarrhea (ie, low plasma protein or low plasma sodium) may also apply to horses affected with diarrhea but with no evidence of Salmonella infection. However, we offer several explanations for the associations observed in the univariable analysis. The association between antimicrobial use and Salmonella shedding has been reported in previous studies. 1, 131,15 6 Admini stration of antibacterial agents, therapeutically or prophylactically, causes disturbances in the ecologic balance between the host and its normal microflora. 17 7 Antimicrobial drugs eliminate intestinal flora that are antagonistic to Salmonella organisms, 1 7 8 giving a competitive advantage to Salmonella and this may explain the observed association between exposure to antimicrobial drugs and Salmonella shedding. The association between anti inflammatory use and Salmonella shedding may be explained by confoun ding factors. Misclassification due to recall bias

PAGE 89

89 by clients or RDVMs is not likely. According to clinicians, most horses with signs of GI disease receive anti inflammatories before admission. The clinical parameters low plasma protein and low plasma sod ium concentration have been documented in diarrheic horses experimentally infected with Salmonella. 17 5 ,17 6 ,17 9 ,1 80 The association between low plasma protein and Salmonella shedding in horses with diarrhea is likely explained by enteric protein loss throug h an impaired gastrointestinal mucosa caused by Salmonella infection. The intestinal damage caused by Salmonella infection results in mucosal disruption, permeability changes, and extensive cell loss especially in the large bowel which facilitates the rele ase of plasma constituents. 17 6 Salmonella enterotoxin interferes with the function of mediators of active colonic secretion including cyclic AMP (cAMP) and prostaglandins E1 and E2, 31,18 1 ,18 2 and disrupts the sodium potassium chloride cotransport system on the basolateral membrane of colonic epithelial cells. 37,18 1 ,18 2 These derangements result in a decrease in sodium absorption, increase in chloride secretion from mucosal cells into the lumen, and secretion of water into the intestinal lumen. 3 7 The finding of high mean corpuscular hemoglobin concentration in horses with diarrhea might be explained by hemoconcentration due to the effect of low sodium or hyposmolality that may occur in horses with diarrhea and/or anorexia. The high plasma triglycerides may be indicative of anorexia caused by Salmonella infection. Salmonella infection in horses can lead to depression anorexia, and hypoglyc emia 17 3 17 6 Model 2: Case horses that tested positive for Salmonella and did not have diarrhea In the multivariable analysis, among case horses without diarrhea that tested positive for Salmonella

PAGE 90

90 plasma protein concentration and high triglyceride concentra tion at admission were associated with shedding. Hyperproteinemia has been reported in Salmonella positive horses with diarrhea, 17 5 but not in Salmonella positive horses without diarrhea. Salmonella infection in horses causes depression, reduction in appet ite and/or anorexia, 17 3 17 6 which may result in inadequate water intake. Deprivation of water can contribute to dehydration manifested as hypovolemia, 18 3 which is associated with several clinical findings including hyperproteinemia. 18 4 18 6 Hyperproteinemia associated with dehydration is characterized by a normal albumin/globulin ratio, because water loss concentrates all plasma proteins proportionally. 18 4 In the present study, the finding of hyperproteinemia was likely attributed to dehydration as all horse s with hyperproteinemia (8 cases and 13 controls) had a normal albumin/globulin ratio. The admission and did not have diarrhea can be explained by anorexia in horses infecte d with Salmonella enterica 17 3 17 6 or a presenting complaint of colic. Model 3: Case horses that tested positive for Salmonella admission and did not have diarrhea In the multivariable analysis, the variables for abdominal surgery and se ason (summer) were retained in the final model. A bdominal surgery can be a risk factor for Salmonella shedding in community acquired infections as shown in this study and in horses with hospital acquired infections 1 32 In this investigation, 16 of 21 case horses tested positive for Salmonella admission and had abdominal surgery The most common Salmonella serotype was Newport (n=4) (one in each year), followed by Saintpaul (n=3), Anatum (n=3), Rubislaw (n=2), and others (n= 9). All 16 case horses were ruled out as nosocomial cases based

PAGE 91

91 on the temporal and spatial distribution of Salmonella cases (which includes serotype, antimicrobial resistance pattern data) identified by the UF LAH surveillance program, as well as the Salmonella environmental contamination status of the hos pital. Horses that undergo abdominal surgery suffer substantial amounts of stress, 122 which may be attributed to insult by the surgical procedure itself and accompanying events such as evacuation and lavage of the large colon or cecum, change or withholdin g of feed, 131 and administration of antimicrobial drugs. 1 60 In humans and mice, surgical stress has been shown to impair the immune system. 16 3 16 5 It is not known if surgical stress (surgical procedure) impairs the immune response system in hospitalized ho rses that undergo abdominal surgery. The accompanying events can cause disturbances in the ecologic balance between the host and its normal microflora 17 7 resulting in alteration of function of normal gastrointestinal flora antagonistic to Salmonella 17 8 In this study, season (summer months) was also associated with Salmonella shedding diarrhea. The association between season and Salmonella shedding is documented, with higher frequen cy of cases reported in summer and fall. 3 15 7 ,18 7 An association between an increase in ambient temperature and risk of nosocomial Salmonella infection has also been described. 131 Increase in ambient temperature may increase the risk of infection by favori ng survival and multiplication of Salmonella organisms in the environment or by compromising immune responses of the host. 131,18 8 In summary, this study makes an attempt to characterize Salmonella shedding in equine GI in patients with or without diarrhea Among horses with diarrhea, results on investigated exposure factors associated with Salmonella shedding are inconclusive

PAGE 92

92 because confounding of exposure factors on Salmonella shedding could not be examined. Among horses without diarrhea, high plasma con centrations of triglycerides at admission, abdominal surgery, and season can be predisposing factors for Salmonella shedding.

PAGE 93

93 Table 3 1. Salmonella serotypes isolated from cases with or without diarrhea Serotype Cases with diarrhea (n=14) Cases without d iarrhea (n=46) Newport 1 13 Saintp aul 0 3 Mbandaka 1 1 Rubislaw 0 3 Anatum 2 6 Braenderup 1 2 Thompson 1 1 Miami 1 1 Cerro 1 0 Uganda 1 0 Gaminara 1 0 Muenster 1 0 Muenchen 2 3 Meleagridis 0 2 Talahassee 0 2 Agona 0 0 Lomalinda 0 1 Other s* 1 8 Rough O:mt: 4,5,12:1: Not known

PAGE 94

94 Table 3 2. Univariable unconditional logistic regression analysis of investigated exposure factors at admission associated with Salmonella after admission in horses with diarrhea Variable Cases with diarrhea n = 14 Controls n = 73 OR* (95% CI)** P value Season Winter and Spring Summer and Fall Prior to admission Antimicrobial use No Yes Anti inflammatory use No Yes Sedative use No Yes Fever No (< 101.5F) Yes (> 101.5F) Leucopenia Yes (< 5.5 K/uL) Nasogastric intubation No Yes Laxative use No Yes Stomach protectant use No Yes Intestinal protectant use No Yes At admission Fever No Yes Leucopenia No Yes 7 7 8 6 4 10 7 7 7 4 2 2 7 7 10 4 13 1 14 0 13 1 7 4 45 28 68 4 5 67 39 33 43 6 1 2 30 40 41 31 68 4 72 0 67 5 59 12 Reference 1. 60 (0. 50 5.07 ) Reference 1 2.75 ( 2.95 55.00 ) Reference 0.18 (0.0 4 0. 81 ) Reference 1.18 (0.37 3. 71 ) Reference 4.09 (0. 91 18. 28 ) Reference ND Reference 0. 75 (0. 23 2. 36 ) Reference 0. 52 (0.1 5 1. 84 ) Reference 1. 30 (0.1 3 1 2.66 ) ND ND ND ND Reference 3.86 (0 92 1 6.19 ) NA 0.41 NA < 0.001 NA 0.02 NA 0.77 NA 0.06 NA ND NA 0. 6 2 NA 0.31 NA 0. 81 ND ND ND ND NA 0.06

PAGE 95

95 Table 3 2. Continued Variable Cases with diarrhea n = 14 Controls n = 73 OR (95% CI) P value At admission Fever and leucopenia No Yes Nasogastri c intubation No Yes Albumin Normal (2.8 4.2 g/dL) Low (< 2.8) High (> 4.2) Creatinine Normal (1.1 2.0 mg/dL) Low (< 1.1) High (> 2.0) Blood urea nitrogen Normal (9 22 mg/dL) Low (< 9 ) High (> 22) Total carbon dioxide Normal (21 34 mEq/L) Low (< 21) High (> 34) Anion gap Normal (10 20 mEq/L) Low (< 10) High (> 20) Alkaline phosphatase Normal (69 228 U/L) Low (< 69) High (> 228) Aspartate aminotransferase Normal (148 322 U/L) Low (< 14 8) High (> 322) Total bilirubin Normal (0.3 1.9 mg/dL) Low (< 0.3) High (> 1.9) Globulin Normal (2.4 4.9 g/dL) Low (< 2.4) High (> 4.9) 14 0 2 12 8 2 0 7 0 4 6 0 4 7 3 0 9 1 1 5 0 5 4 1 5 2 0 8 8 0 1 73 0 8 62 60 3 4 45 3 19 49 1 17 56 8 2 53 4 9 46 0 21 33 1 32 16 0 51 55 3 9 ND ND Reference 0.77 (0.14, 4. 1 0) Reference 5.00 ( 0.72 34.63 ) ND Reference ND 1. 70 (0. 42 6.75 ) Reference ND 1. 92 (0.4 8 7.63 ) Reference 3.00 (0. 64 14.02 ) ND Reference 2.33 (0. 2 1, 25.24 ) 0. 87 (0.0 9 7.95 ) Reference ND 2. 19 (0. 57 8. 3 8) Reference 5.00 (0. 28 88.53 ) 0. 7 8 (0.2 3 2. 62 ) Reference ND 1. 25 (0.24 6.52 ) Reference ND 0. 7 6 (0.0 8 6.86 ) ND ND NA 0.76 NA 0.10 ND NA ND 0.44 NA ND 0. 35 NA 0.16 ND NA 0. 48 0. 90 ND ND 0.25 NA 0.27 0. 68 NA ND 0. 7 8 NA ND 0. 81

PAGE 96

96 Table 3 2. Continued Variable Cases with diarrhea n = 14 Controls n = 73 OR (95% CI) P value At admission Albumin/Globulin ratio Normal (0.6 1.4) Low (< 0.6) High (>1.4) Glucose Normal (62 128 mg/dL) Low (< 62) High (> 128) Triglycerides Normal (5 45 mg/dL) Low (< 5) High (> 45) Magnesium Normal (1.3 1.9 mg/dL) Low (< 1.3) High (> 1.9) Gamma gl utamyl transferase Normal (17 50 U/L) Low (< 17) High (> 50) Creatine kinase Normal (134 729 U/L) Low (< 134) High (> 729) Sodium Normal (134 143 mEq/L) Low (< 134) High (> 143) Potassium Normal (2.2 5.3 mEq/L) Low (< 2.2) High (> 5.3) Chloride Normal (96 105 mEq/L) Low (< 96) High (> 105) Calcium Normal (10.9 13.4 mg/dL) Low (< 10.9) High (> 13.4) Color o f mucus membranes Pink Icteric/yellow Red/injected/toxic line 9 0 0 4 0 6 4 0 7 7 3 0 8 0 2 8 0 2 2 8 1 6 4 0 4 6 0 7 3 0 8 1 4 63 2 2 34 1 31 52 0 14 51 12 4 44 4 19 38 1 28 50 15 1 67 0 0 44 23 0 49 15 2 49 1 23 ND ND ND Reference ND 1. 64 (0 .4 2 6. 38 ) Reference ND 5.67 (1. 40 22.92 ) Reference 1. 82 (0. 40 8.09 ) ND Reference ND 0. 57 (0. 11 2. 9 8) Reference ND 0.3 3 (0.0 6 1. 72 ) Reference 13.6 (2. 60 71.02 ) ND ND ND ND Reference 2. 86 (0. 73 11.20 ) ND Reference 1. 40 (0. 32 6.09 ) ND Referen ce 5.44 (0. 31 95.21 ) 0.9 4 (0.26, 3.3 9 ) ND ND ND NA ND 0.47 NA ND 0.0 1 NA 0. 43 ND NA ND 0. 51 NA ND 0.19 NA 0.002 ND ND ND ND NA 0. 1 2 ND NA 0. 65 ND NA 0. 24 0.93

PAGE 97

97 Table 3 2. Continued Variable Cases with diarrhea n = 14 Controls n = 73 OR (95% C I) P value At admission Capillary refill time Abnormal (> 2) Heart rate Normal (28 40 beats/min) Elevated (41 60) High (> 60) Lactate (blood gas) Abnormal (> 1.5) Plasma protein Normal (5.9 7.9 g/dL) Low (< 5.9) High (> 7.9) Pa cked cell volume Normal (30 48%) Low (< 30) High (> 48) White blood cell count Normal (5.5 11.0 K/uL) Low (< 5.5) High (> 11.0) Fibrinogen Normal (100 400 mg/dL) Low (< 100) High (> 400) Percent Neutrophils Normal (28 83%) Lo w (< 28) High (> 83) Percent Monocytes Normal (1 11%) Low (< 1) High (> 11) Toxicity No Yes Percent lymphocytes Normal (20 59%) Low (< 20) High (> 59) 8 5 2 8 4 5 5 5 7 1 7 1 5 5 4 2 7 0 3 7 1 2 7 0 3 4 6 3 7 1 50 23 12 47 14 26 33 44 10 10 43 9 13 45 12 14 53 0 13 51 2 15 65 0 2 43 21 23 44 0 Reference 1.3 5 (0.40, 4. 60 ) Reference 1.02 (0.1 9 5.44 ) 1. 71 (0.2 6 11.05 ) Reference 0. 78 (0 20 3.01 ) Reference 6.18 (1.7 1 2 2.30 ) 0. 88 (0.09, 8 .15) Reference 0.7 0 (0.0 7 6.36 ) 2. 45 (0.6 8 8.75 ) Reference 2. 26 (0. 58 8.82 ) 0.47 (0. 05 4.09 ) Reference ND 1.7 4 (0. 39 7. 69 ) Reference 2. 54 (0.2 1 30.57 ) 0.6 7 (0.1 3 3. 3 9) Reference ND 1 3.92 (1. 97 98.06 ) Reference 3. 07 (0. 7 8, 1 2.06 ) Reference 1.86 (0. 43 7.92 ) ND NA 0.62 NA 0.98 0.57 NA 0.72 NA 0.005 0.91 NA 0. 75 0.16 NA 0. 23 0. 4 9 NA ND 0.46 NA 0.46 0.63 NA ND 0.0 08 NA 0. 1 0 NA 0.39 ND

PAGE 98

98 Table 3 2. Continued Variable Cases with diar rhea n = 14 Controls n = 73 OR (95% CI) P value At admission Neutrophils Normal (2.2 8.1 K/uL) Low (< 2.2) High (>8.1) Neutropenia 2.2 K/uL) Yes (< 2.2) RBC Normal (6.7 10 M/uL) Low (< 6.7) High (> 10) Hemoglobin Normal (11.2 16.2 g/dL) Low (< 11.2) High (> 16.2) Hematocrit (calculated) Normal (30 43%) Low (< 30) High (> 43) Mean corpuscular volume N ormal (37.5 50.0 fL) Low (< 37.5) High (> 50.0) Mean corpuscular hemoglobin Normal (14 18.7 pg) Low (< 14) High (> 18.7) Mean corpuscular hemoglobin concentration Normal (36.4 38.8 g/dL) Low (< 36.4) High (> 38.8) Corpuscular h emog lobin concentration mean Normal (34.9 37.6 g/dL) Low (< 34.9) High (> 37.6) Corpuscular hemoglobin Normal (13.7 17.9 pg) Low (< 13.7) High (>17.9) 7 3 1 8 3 5 1 3 5 1 4 6 1 3 9 1 0 8 1 1 6 0 5 3 0 7 5 1 4 42 8 21 63 8 52 9 7 51 6 11 45 10 12 61 2 3 63 2 3 52 7 12 46 1 21 58 1 9 Reference 2.39 (0.50, 11.37) 0.26 (0.03, 2.30) Reference 3.17 (0.68, 14.64) Reference 0.6 3 (0.0 7 5.56 ) 2.4 4 ( 0.53 11. 06 ) Reference 1. 03 (0.1 1 9.65 ) 2. 26 (0.5 9 8.67 ) Reference 0.5 1 (0.05, 4. 44 ) 1. 27 (0. 30 5. 35 ) Reference 3.38 (0. 27 41.30 ) ND Reference 2.83 (0.2 3 3 3.75 ) 1. 88 (0.18 1 9.70 ) Reference ND 4.54 ( 1.11 1 8.45 ) Reference ND 5.11 ( 1.20 21.73 ) Reference 7.00 (0. 40 122.04 ) 3.11 (0. 79 1 2.23 ) NA 0.27 0.22 NA 0.13 ND 0.68 0.24 NA 0.97 0.23 NA 0.54 0.74 NA 0.33 ND NA 0.41 0. 59 NA ND 0.03 NA ND 0.02 NA 0.18 0.10

PAGE 99

99 Table 3 2. Continued Variable Cases with diarrhea n = 14 Controls n = 73 OR (95% CI) P value At admission Red cell distribution width Normal (16.3 19%) Low (< 16.3) High (> 19) Hemoglobin distribution width Normal (1.5 2.1 g/dL) Low (< 1.5) High (> 2.1) Platelet count Normal (100 250 K/uL) Low (< 100) High (> 250) Mean platelet volume Nor mal (5.6 10.4 fL) Low (< 5.6) High (> 10.4) Icterus Normal (5 25 Units) Low (< 5) High (> 25) Year of admission 2007 2008 2009 2010 6 1 3 4 0 6 8 2 0 9 0 1 5 0 5 3 6 2 3 57 0 11 37 0 30 60 3 3 64 0 3 50 1 14 20 22 18 13 Reference ND 2.59 (0.56, 11.95) Reference ND 1. 85 (0. 47 7.16 ) Reference 5.00 (0. 7 2, 34 63 ) ND Reference ND 2.37 (0.2 2 25.31 ) Reference ND 3. 57 (0. 90 1 4.11 ) Reference 1.81 (0.40, 8.25) 0.74 (0.11, 4.94) 1.53 (0.26, 8.81) NA ND 0 .22 NA ND 0. 37 NA 0.10 ND NA ND 0.47 NA ND 0.06 NA 0.43 0.75 0.62 OR = crude odds ratios; ** 95% CI = 95% confidence interval; NA = Not applicable; ND = Not determined

PAGE 100

100 Table 3 3. Univariable unconditional logistic regression analysis of investigat ed exposure factors at admission associated with Salmonella after admission in horses without diarrhea Variable Cases with diarrhea n = 25 Controls n = 73 OR* (95% CI)** P value Season Winter and Spring Summer and Fall Prior to admission Antimicrobial use No Yes Anti inflammatory use No Yes Sedative use No Yes Fever No (< 101.5F) Yes (> 101.5F) Leucopenia Yes (< 5.5 K/uL) Nasogastric intubation No Yes Laxative use No Yes Stomach protectant use No Yes Intestinal protectant use No Yes At admission Fever No Yes Leucopenia No Yes 14 11 23 1 4 20 18 6 11 3 2 3 9 `13 14 9 21 1 22 0 20 4 15 7 45 28 68 4 5 67 39 33 43 6 1 2 30 40 41 31 68 4 72 0 67 5 59 12 Reference 1.2 6 (0. 50 3.16) Reference 0. 73 (0.0 7 6.95 ) Reference 0. 37 (0. 09 1. 52 ) Reference 0. 39 (0.1 4 1. 10 ) Reference 1. 95 (0. 42 9.08 ) Reference 0.75 (0.0 3 14.97 ) Reference 1.0 8 (0. 40 2.8 6 ) Reference 0. 85 (0.32 2. 21 ) Reference 0. 80 (0.0 8 7.64 ) ND ND Reference 1.54 (0. 26 9.02 ) Reference 2.29 ( 0.77 6.83 ) NA 0.61 NA 0.79 NA 0.16 NA 0.07 NA 0. 39 NA 0.85 NA 0.87 NA 0.74 NA 0.85 ND ND NA 0.62 NA 0. 13

PAGE 101

101 Table 3 3. Continued Variable Cases with diarrhea n = 25 Contro ls n = 73 OR (95% CI) P value At admission Fever and leucopenia No Yes Nasogastric intubation No Yes Albumin Normal (2.8 4.2 g/dL) Low (< 2.8) High (> 4.2) Creatinine Normal (1.1 2.0 mg/dL) Low (< 1.1) High (> 2.0) Blood urea ni trogen Normal (9 22 mg/dL) Low (< 9 ) High (> 22) Total carbondioxide Normal (21 34 mEq/L) Low (< 21) High (> 34) Anion gap Normal (10 20 mEq/L) Low (< 10) High (> 20) Alkaline phosphatase Normal (69 228 U/L) Low (< 69) High (> 228) Aspartate aminotransferase Normal (148 322 U/L) Low (< 148) High (> 322) Total bilirubin Normal (0.3 1.9 mg/dL) Low (< 0.3) High (> 1.9) Globulin Normal (2.4 4.9 g/dL) Low (< 2.4) High (> 4.9) 22 0 3 22 18 2 1 1 2 0 10 11 0 10 15 4 1 18 2 2 14 0 7 9 0 12 7 0 14 18 1 1 70 0 8 62 60 3 4 45 3 19 49 1 17 56 8 2 44 3 8 46 0 21 33 1 32 16 0 51 55 3 9 ND ND Reference 0. 94 (0. 23 3. 88 ) Refere nce 2.00 (0. 31 12. 75 ) 0.7 5 (0.07, 7. 07 ) Reference ND 1.97 (0. 72 5.34 ) Reference ND 2. 62 (0. 94 7.25 ) Reference 1.57 (0. 42 5.78 ) 1.5 7 (0.1 3 1 8.30 ) Reference 1.62 (0. 25 10.58 ) 0.61 (0. 11 3.16 ) Reference ND 1. 09 (0.3 8 3. 11 ) Reference ND 1. 37 (0. 5 1, 3. 70 ) Reference ND 0. 6 2 (0.2 1 1.82 ) Reference 0. 88 (0.0 8 8.93 ) 0. 29 (0.0 3 2.45 ) ND ND NA 0.93 NA 0.46 0. 80 NA ND 0. 1 8 NA ND 0.06 NA 0. 49 0.7 1 NA 0. 60 0. 55 NA ND 0. 86 NA ND 0. 52 NA ND 0. 39 NA 0.91 0. 25

PAGE 102

102 Table 3 3. Continued Variable Cas es with diarrhea n = 25 Controls n = 73 OR (95% CI) P value At admission Albumin/Globulin ratio Normal (0.6 1.4) Low (< 0.6) High (>1.4) Glucose Normal (62 128 mg/dL) Low (< 62) High (> 128) Triglycerides Normal (5 45 mg/dL) Low (< 5) High (> 45) Magnesium Normal (1.3 1.9 mg/dL) Low (< 1.3) High (> 1.9) Gamma glutamyl transferase Normal (17 50 U/L) Low (< 17) High (> 50) Creatine kinase Normal (134 729 U/L) Low (< 134) High (> 729) Sodium Normal (134 143 mE q/L) Low (< 134) High (> 143) Potassium Normal (2.2 5.3 mEq/L) Low (< 2.2) High (> 5.3) Chloride Normal (96 105 mEq/L) Low (< 96) High (> 105) Calcium Normal (10.9 13.4 mg/dL) Low (< 10.9) High (> 13.4) 20 0 1 9 0 12 12 0 10 18 1 2 14 4 3 11 0 10 13 9 0 21 0 0 15 6 0 15 6 0 63 2 2 34 1 31 52 0 14 51 12 4 44 4 19 38 1 28 50 15 1 67 0 0 44 23 0 49 15 2 Reference ND 1.57 (0. 13 1 8.30 ) Reference ND 1.46 (0.5 4 3.9 4 ) Reference ND 3.09 ( 1.10 8. 63 ) Reference 0. 21 (0.0 2 1.76 ) 1. 29 (0.2 2 7.58 ) Reference 2. 77 (0. 6 2, 12.28 ) 0. 43 (0.1 1 1.66 ) Reference ND 1. 23 (0.4 6 3. 30 ) Reference 2.30 ( 0.82 6.44 ) ND ND ND ND Reference 0.76 (0. 26 2.23 ) ND Reference 1. 30 (0. 4 3, 3. 96 ) ND NA ND 0. 71 NA ND 0.45 NA ND 0.03 NA 0.15 0.77 NA 0.17 0.22 NA ND 0.67 NA 0. 11 ND ND ND ND NA 0.62 ND NA 0.63 ND

PAGE 103

103 Table 3 3. Continued Variable Cases with diarrhea n = 25 Controls n = 73 OR (95% CI) P value At admission Co lor of mucus membranes Pink Icteric/yellow Red/injected/toxic line Capillary refill time Abnormal (> 2) Heart rate Normal (28 40 beats/min) Elevated (41 60) High (> 60) Lactate (blood gas) Abnormal (> 1.5) White blood cell count Normal (5.5 11.0 K/uL) Low (< 5.5) High (> 11. 0) Packed cell volume Normal (30 43%) Low (< 30) High (> 43) Plasma protein Normal (5.9 7.9 g/dL) Low (< 5.9) High (> 7.9) Fibrinogen Normal (100 400 mg/dL) Low (< 100) High (> 400) Percent Neutrophils Normal (28 83%) Low (< 28) High (> 83) Percent Monocytes Normal (1 11%) Low (< 1) High (> 11) Toxicity No Yes 14 0 11 16 9 3 16 6 11 11 15 5 3 12 0 8 12 2 8 15 1 4 17 1 3 17 0 4 11 9 49 1 23 50 23 12 47 14 26 33 35 11 13 48 6 13 52 6 13 53 0 13 51 2 15 65 0 2 43 21 Reference ND 1. 67 (0. 65 4.25 ) Reference 1.2 2 (0.4 7 3. 1 7) Reference 1. 36 (0.3 4 5.44 ) 1.71 (0. 35 8.37 ) Reference 0. 78 (0. 29 2.1 0 ) Reference 1.06 ( 0.31 3.58 ) 0.53 (0. 13 2.16 ) Reference ND 2. 46 (0. 83 7.28 ) Reference 1. 44 (0.2 5 8.05 ) 2.66 (0.9 0 7.86 ) Reference ND 1.08 (0. 30 3. 82 ) Reference 1. 50 (0.12 1 7.60 ) 0.6 0 (0.15 2. 32 ) Reference ND 7.64 (1. 29 45.31 ) Reference 1.6 7 (0. 60 4. 66 ) NA ND 0. 2 7 NA 0.67 NA 0.6 6 0. 50 NA 0.63 NA 0. 92 0. 38 NA ND 0.10 NA 0. 67 0.0 7 NA ND 0. 8 9 NA 0.74 0. 46 NA ND 0.02 NA 0.3 2

PAGE 104

104 Table 3 3. Continued Variable Cases with diarrhea n = 25 Controls n = 73 OR (95% CI) P value At admission Percent lymphocytes Normal (20 59%) Low (< 2 0) High (> 59) Neutrophils Normal (2.2 8.1 K/uL) Low (< 2.2) High (>8.1) Neutro penia Yes (< 2.2) RBC Normal (6.7 10 M/uL) Low (< 6.7) High (> 10) Hemoglobin Normal (11.2 16.2 g/dL) Low (< 11.2) High (> 16.2) Hematocrit (calculated ) Normal (30 43%) Low (< 30) High (> 43) Mean corpuscular volume N ormal (37.5 50.0 fL) Low (< 37.5) High (> 50.0) Mean corpuscular hemoglobin Normal (14 18.7 pg) Low (< 14) High (> 18.7) Mean corpuscular hemoglobin concentration Normal (36.4 38.8 g/dL) Low (< 36.4) High (> 38.8) Corpuscular h emogl obin concentration mean Normal (34.9 37.6 g/dL) Low (< 34.9) High (> 37.6) 9 12 0 15 4 4 16 6 14 1 6 15 0 6 15 0 6 20 1 0 19 1 1 16 4 1 16 0 5 23 44 0 34 7 18 63 8 52 9 7 51 6 11 45 10 12 61 2 3 63 2 3 50 7 11 47 1 20 Reference 0.6 9 (0.25, 1.89) ND Reference 1. 29 (0. 3 2, 5.0 9 ) 0. 50 (0.14 1.74 ) Reference 2.9 5 (0.8 9 9.72 ) Reference 0. 35 (0.04, 2.96 ) 2. 7 1 (0.80, 9. 12 ) Reference ND 1.85 (0. 58 5.85 ) Referenc e ND 1. 50 (0. 4 7, 4.69 ) Reference 1. 5 2 (0.13 1 7.72 ) ND Reference 1. 47 (0.1 2 1 7.07 ) 0.98 (0. 09 9.94 ) Reference 1. 57 (0. 41 5.94 ) 0.2 5 (0.03, 2. 06 ) Reference ND 1.12 (0.32, 3.96) NA 0.47 ND NA 0. 71 0. 27 NA 0.07 NA 0.3 3 0.10 NA ND 0. 29 NA ND 0 48 NA 0.7 3 ND NA 0. 75 0.99 NA 0. 50 0.19 NA ND 0.85

PAGE 105

105 Table 3 3. Continued Variable Cases with diarrhea n = 25 Controls n = 73 OR (95% CI) P value At admission Corpuscular hemoglobin Normal (13.7 17.9 pg) Low (< 13.7) High (>17.9) Red cell d istribution width Normal (16.3 19%) Low (< 16.3) High (> 19) Hemoglobin distribution width Normal (1.5 2.1 g/dL) Low (< 1.5) High (> 2.1) Platelet count Normal (100 250 K/uL) Low (< 100) High (> 250) Mean platelet volume Normal (5 .6 10.4 fL) Low (< 5.6) High (> 10.4) Icterus Normal (5 25 Units) Low (< 5) High (> 25) Year of admission 2007 2008 2009 2010 16 1 4 16 0 5 13 0 8 19 1 0 21 0 0 14 1 5 5 3 11 6 58 1 9 57 0 11 37 0 30 60 3 3 64 0 3 50 1 14 20 22 18 13 Reference 3.15 (0.1 8 52.69 ) 1. 40 (0. 38 5.03 ) Reference ND 1. 61 (0.4 9 5.34 ) Reference ND 0. 75 (0.2 7 2. 07 ) Reference 1.05 (0. 1 0, 10.72 ) ND ND ND ND Reference 3. 05 (0.1 8 51.20 ) 1.0 9 (0.3 4 3. 42 ) Reference 0.54 (0.11, 2.58) 2.44 (0.71, 8.39) 1.84 (0.46, 7.31) NA 0. 42 0. 60 NA ND 0. 42 NA ND 0. 59 NA 0. 96 ND ND ND ND NA 0.4 3 0.88 NA 0.44 0.15 0.38 OR = crude odds ratios; ** 95% CI = 95% confidence interval; NA = Not applicabl e; ND = Not determined

PAGE 106

106 Table 3 4. Univariable unconditional logistic regression analysis of investigated exposure factors (during hospitalization on days 2 and 3) associated with Salmonella shedding Variable Cases with diarrhea n = 21 Controls n = 73 OR* (95% CI)** P value Season Winter and Spring Summer and Fall Prior to admission Antimicrobial use No Yes Anti inflammatory use No Yes Sedative use No Yes Fever No (< 101.5F) Yes (> 101.5F) Leucopenia Yes (< 5.5 K/uL) Nasogastric intubation No Yes Laxative use No Yes Stomach protectant use No Yes Intestinal protectant use No Yes A t admission Fever No Yes Leucopenia No Yes 7 14 21 0 1 20 10 11 15 1 0 0 10 11 9 12 18 3 21 0 20 1 19 0 45 28 68 4 5 67 39 33 43 6 1 2 30 40 41 31 68 4 72 0 67 5 59 12 Refer ence 3. 21 (1. 15 8. 93 ) ND ND Reference 1. 49 (0.1 6 1 3.52 ) Reference 1. 30 (0.4 9 3. 44 ) Reference 0. 47 (0.0 5 4.29 ) ND ND Reference 0. 82 (0. 31 2. 19 ) Reference 1.7 6 (0.66, 4.7 0 ) Reference 2.83 (0. 58 1 3.81 ) ND ND ND ND ND ND NA 0.02 ND ND NA 0.72 NA 0.59 NA 0.5 1 ND ND NA 0. 70 NA 0.25 NA 0.19 ND ND ND ND ND ND

PAGE 107

107 Table 3 4. Continued Variable Cases with diarrhea n = 21 Controls n = 73 OR (95% CI) P value At admission Fever and leucopenia No Yes Nasogastric intubation No Y es Albumin Normal (2.8 4.2 g/dL) Low (< 2.8) High (> 4.2) Creatinine Normal (1.1 2.0 mg/dL) Low (< 1.1) High (> 2.0) Blood urea nitrogen Normal (9 22 mg/dL) Low (< 9 ) High (> 22) Total carbondioxide Normal (21 34 mEq/L) Low (< 21) High (> 34) Anion gap Normal (10 20 mEq/L) Low (< 10) High (> 20) Alkaline phosphatase Normal (69 228 U/L) Low (< 69) High (> 228) Aspartate aminotransferase Normal (148 322 U/L) Low (< 148) High (> 322) Total bilirubin Normal (0.3 1.9 mg/dL) Low (< 0.3) High (> 1.9) Globulin Normal (2.4 4.9 g/dL) Low (< 2.4) High (> 4.9) 19 0 1 18 18 1 0 14 0 5 15 0 4 16 3 0 16 1 2 17 0 2 13 0 6 7 1 11 17 2 0 70 0 8 62 60 3 4 45 3 19 49 1 17 56 8 2 53 4 9 46 0 21 33 1 32 16 0 51 55 3 9 ND ND Reference 2. 32 (0. 27 19.82 ) Reference 1.1 1 (0.1 0 11. 34 ) ND Reference ND 0. 9 1 (0.2 8 2. 90 ) Reference ND 0.7 6 (0.22, 2. 63 ) Reference 1. 3 1 (0. 31 5.53 ) ND Reference 1.14 (0. 11 11.72 ) 0.8 6 (0.1 6 4. 42 ) Reference ND 0.2 5 (0.05, 1.2 1 ) Reference ND 0.4 7 (0.16, 1.4 0 ) Reference ND 0.4 9 (0.16, 1.4 8 ) Reference 2. 15 (0.3 3 1 3.99 ) ND ND ND NA 0.44 NA 0.92 ND NA ND 0.87 NA ND 0.6 7 NA 0. 71 ND NA 0. 90 0.85 NA ND 0.08 NA ND 0.17 NA ND 0.20 NA 0. 42 ND

PAGE 108

108 Table 3 4. Continued Variable Cases with diarrhea n = 21 Controls n = 73 OR (95% CI) P value At admission Albumin/Globulin ratio Normal (0.6 1.4) Low (< 0.6) High (>1.4) Glucose Normal (62 128 mg/dL) Low (< 62) High (> 128) Triglycerides Normal (5 45 mg/dL) Low (< 5) High (> 45) Magnesium Normal (1.3 1.9 mg/dL) Low (< 1.3) High (> 1.9) Gamma glutamyl transferase Normal (17 50 U/L) Low (< 17) High (> 5 0) Creatine kinase Normal (134 729 U/L) Low (< 134) High (> 729) Sodium Normal (134 143 mEq/L) Low (< 134) High (> 143) Potassium Normal (2.2 5.3 mEq/L) Low (< 2.2) High (> 5.3) Chloride Normal (96 105 mEq/L) Low (< 96) High (> 105) Calcium Normal (10.9 13.4 mg/dL) Low (< 10.9) High (> 13.4) Color o f mucus membranes Pink Icteric/yellow Red/injected/toxic line 17 0 2 11 0 8 17 1 1 11 6 2 14 2 3 15 0 4 14 5 0 19 0 0 8 11 0 11 8 0 15 0 6 63 2 2 34 1 31 52 0 14 51 12 4 44 4 19 38 1 28 50 15 1 67 0 0 44 23 0 49 15 2 49 1 23 Reference ND 3. 70 (0.4 8 28. 27 ) Reference ND 0.7 9 (0.2 8 2. 24 ) Reference ND 0.2 2 (0.02, 1.8 2 ) Referenc e 2. 19 (0. 69 6.92 ) 2.19 (0.36 1 3.27 ) Reference 1. 56 (0.2 6 9.34 ) 0.4 9 (0.1 2 1. 88 ) Reference ND 0.3 6 (0.1 0 1.2 0 ) Reference 1. 21 (0.3 7 3. 92 ) ND ND ND ND Reference 2. 6 3 (0.9 2, 7.44 ) ND Reference 2. 3 7 (0.8 0 6.98 ) ND Reference ND 0.8 5 (0.2 9 2. 48 ) NA ND 0.20 NA ND 0.66 NA ND 0.16 NA 0.18 0. 39 NA 0. 62 0. 30 NA ND 0.09 NA 0. 74 ND ND ND ND NA 0.06 ND NA 0.11 ND NA ND 0.76

PAGE 109

109 Table 3 4. Continued Variable Cases with diarrhea n = 21 Controls n = 73 OR (95% CI) P value At admission Capillary re fill time Abnormal (> 2) Heart rate Normal (28 40 beats/min) Elevated (41 60) High (> 60) Lactate (blood gas) Abnormal (> 1.5) Plasma protein Normal (5.9 7.9 g/dL) Low (< 5.9) High (> 7.9) Packed cell v olume Normal (30 48%) Low (< 30) High (> 48) White blood cell count Normal (5.5 11.0 K/uL) Low (< 5.5) High (> 11.0) Fibrinogen Normal (100 400 mg/dL) Low (< 100) High (> 400) Percent Neutrophils Normal (28 83%) Low (< 28) Hig h (> 83) Percent Monocytes Normal (1 11%) Low (< 1) High (> 11) Toxicity No Yes Percent lymphocytes Normal (20 59%) Low (< 20) High (> 59) 13 8 7 10 4 7 10 17 3 0 17 1 2 15 0 4 18 1 0 12 0 7 18 0 1 9 7 5 14 0 50 23 12 47 14 26 33 44 10 10 43 9 13 45 12 14 53 0 13 51 2 15 65 0 2 43 21 23 44 0 Reference 1. 33 (0. 48 3.67 ) Reference 0.3 6 (0.11, 1.15 ) 0. 48 (0.11, 2. 08 ) Reference 1. 12 (0.3 7 3. 36 ) Reference 0. 7 7 (0.1 9 3.16 ) ND Reference 0.3 1 (0.03, 2.6 6 ) 0.4 3 (0.0 8 2. 12 ) Reference ND 1.02 (0.2 8 3.65 ) ND ND ND Reference ND 1. 98 (0. 66 5. 93 ) Reference ND 1. 80 (0.1 5 21. 0 6) Reference 1. 5 9 (0.5 2 4.86 ) Reference 1. 46 (0.4 6 4. 5 7) ND NA 0. 57 NA 0.08 0.33 NA 0.83 NA 0. 72 ND NA 0.28 0.30 NA ND 0.9 7 ND ND ND NA ND 0. 22 NA ND 0.63 NA 0. 41 NA 0. 51 ND

PAGE 110

110 Table 3 4. Continued Variable Cases with diarrhea n = 21 Controls n = 73 OR (95% CI) P value At admission Neutr ophils Normal (2.2 8.1 K/uL) Low (< 2.2) High (>8.1) Neutropenia Yes (< 2.2) RBC Normal (6.7 10 M/uL) Low (< 6.7) High (> 10) Hemoglobin Normal (11.2 16.2 g/dL) Low (< 11.2) High (> 16.2) Hematocrit (calculated) Normal (30 43%) Low (< 30) High (> 43) Mean corpuscular volume N ormal (37.5 50.0 fL) Low (< 37.5) High (> 50.0) Mean corpuscular hemoglobin Normal (14 18.7 pg) Low (< 14) High (> 18.7) Mean corpuscular hemoglobin concentration Normal (36.4 38.8 g/dL) Low (< 36.4) High (> 38.8) Co rpuscular Hemogl obin concentration mean Normal (34.9 37.6 g/dL) Low (< 34.9) High (> 37.6) Corpuscular hemoglobin Normal (13.7 17.9 pg) Low (< 13.7) High (>17.9) 14 0 5 19 0 14 4 1 14 2 3 14 3 2 16 1 2 13 1 5 13 0 6 9 0 10 9 2 8 42 8 21 63 8 52 9 7 51 6 11 45 10 12 61 2 3 63 2 3 50 7 11 47 1 20 58 1 9 Reference ND 0.71 (0.22, 2.29) ND ND Reference 1.5 8 (0.4 3 5. 82 ) 0. 50 (0.05, 4. 4 4) Reference 1. 16 (0.2 1 6.34 ) 0. 9 5 (0.23, 3. 84 ) Reference 0.9 5 (0.2 3 3.90) 0.5 3 (0.1 0 2.6 2 ) Reference 1. 90 (0.16, 2 2.37 ) 2.54 (0.3 9 1 6.52 ) Reference 2.2 6 (0. 19 2 6.65 ) 7.55 (1. 62 3 5.12 ) Reference ND 2.09 (0. 65 6. 73 ) Reference ND 2. 61 (0. 92 7.39 ) Reference 11.45 (0. 95 1 37. 39 ) 5.09 (1. 61 1 6.04 ) NA ND 0.57 ND ND NA 0.4 8 0. 5 4 NA 0. 85 0.94 NA 0.95 0.43 NA 0.6 0 0.32 NA 0.51 0.009 NA ND 0.21 NA ND 0.07 NA 0.05 0.005

PAGE 111

111 Table 3 4. Continued Variable Cases with diarrhea n = 21 Controls n = 73 OR (95% CI) P value At a dmission Red cell distribution width Normal (16.3 19%) Low (< 16.3) High (> 19) Hemoglobin distribution width Normal (1.5 2.1 g/dL) Low (< 1.5) High (> 2.1) Platelet count Normal (100 250 K/uL) Low (< 100) High (> 250) Mean platelet volume Normal (5.6 10.4 fL) Low (< 5.6) High (> 10.4) Icterus Normal (5 25 Units) Low (< 5) High (> 25) During hospitalization Antimicrobial use No Yes Anti inflammatory use No Yes Sedative use No Yes Fever (day 2 or 3) No Yes Leucopenia No Yes Fever and leucopenia No Yes Nasogastric intubation No Yes 15 0 4 12 0 7 16 2 1 18 0 1 14 1 4 3 18 1 20 0 21 16 5 1 5 3 3 10 11 57 0 11 37 0 30 60 3 3 64 0 3 50 1 14 29 44 9 64 1 72 45 28 12 6 16 2 34 37 Reference ND 1.38 (0.38, 4.95) Reference ND 0. 7 1 (0.2 5 2.05 ) Reference 2. 50 (0.3 8 1 6.25 ) 1. 25 (0.1 2 1 2.84 ) Reference ND 1. 18 (0.11, 1 2.09 ) Reference 3.62 (0. 21 61.21 ) 1.03 (0.2 9 3. 58 ) Reference 3. 95 ( 1.06 1 4.64 ) Reference 2. 81 (0.3 3 2 3.57 ) ND ND Reference 0. 50 (0.16, 1.5 2 ) Reference 10.00 (0. 9 4, 105.92 ) Reference 8.00 (0. 91 70.27 ) Reference 1.0 1 (0.3 8 2. 6 7) NA ND 0.61 NA ND 0. 53 NA 0.33 0. 85 NA ND 0.88 NA 0. 37 0.95 NA 0.03 NA 0.34 ND ND NA 0.22 NA 0.05 NA 0.06 NA 0. 98

PAGE 112

112 Table 3 4. Continued Variable Cases with diarrhea n = 21 Controls n = 73 OR (95% CI) P value During hospitalization Feed withholding No Yes Laxative use No Yes Stomach pr otectant use No Yes Intestinal protectant use No Yes General anesthesia No Yes Abdominal surgery No Yes Location of surgical lesion Large intestine Small intestine Small & large intestine Type of surgical lesion Strangulating Non strangulating Antispasmodic use No Yes Plasma protein (day 2) Normal (5.9 7.9 g/dL) Low (< 5.9) High (> 7.9) Plasma protein (day 3) Normal (5.9 7.9 g/dL) Low (< 5.9) High (> 7.9) Packed cell volume (day 2) Normal (30 48) Low (< 30) High (> 48) 5 16 13 7 13 8 20 1 5 16 5 16 4 11 1 2 14 8 13 6 13 0 9 7 0 14 4 1 7 66 50 23 63 10 70 3 35 38 35 38 5 28 3 5 31 33 40 36 25 2 30 16 0 48 9 6 Reference 0.33 ( 0.09, 1.20) Reference 1. 17 (0.4 1 3.32 ) Reference 3.87 (1. 28 1 1.70 ) Reference 1. 16 (0.1 1 1 1.83 ) Reference 2. 94 (0.9 7 8. 89 ) Reference 2. 94 (0.9 7 8. 89 ) Reference 0. 49 (0.11, 2. 1 7) 0.4 1 (0.0 3 5.70 ) Reference 1. 12 (0.1 9 6. 54 ) Reference 1. 34 (0.4 9 3. 62 ) Reference 3. 12 ( 1.04 9. 31 ) ND Reference 1. 4 5 (0. 45 4. 6 4) ND Reference 1.5 2 (0.4 0 5. 70 ) 0.5 7 (0.06, 5.15 ) NA 0.09 NA 0.76 NA 0.01 NA 0.89 NA 0.05 NA 0.05 NA 0.34 0.5 1 NA 0. 8 9 NA 0. 5 6 NA 0.04 ND NA 0. 5 2 ND NA 0.53 0. 61

PAGE 113

113 Table 3 4. Continued Variable Cases with diarrhea n = 21 Controls n = 73 OR (95% CI) P value During hospitalization Packed cell volume (day 3) Normal (30 48) Low (< 30) High (> 48) Color of mucus membranes (day 2) Pink Icteric/yellow Red/injected/toxic line Color of mucus membranes (day 3) Pink Icteric/yellow Red/injected/toxic line Capillary refill time (day 2) Abnormal (> 2) Capillary refill time (day 3) Abnormal (> 2s) Heart rate (day 2) Normal (28 40 beats per min) Elevated (41 60) High (> 60) Heart rate (day 3) Normal (28 40 beats per min) Elevated (41 60) High (> 60) Year of admission 2007 2008 2009 2010 Number of fecal samples tested 13 2 0 16 0 5 18 0 3 16 5 16 5 3 14 4 6 11 4 4 8 6 3 37 8 1 55 5 13 64 3 5 50 23 57 13 14 45 14 24 42 3 20 22 18 13 Reference 0.71 (0.13, 3.79) ND Reference ND 1.3 2 (0.40, 4.2 6 ) Reference ND 2. 1 3 (0. 46 9.79 ) Reference 0.6 7 (0.22, 2.0 8 ) Reference 1.3 7 (0.42 4.4 1 ) Reference 1. 45 (0.3 6 5. 79 ) 1. 33 (0.2 5 7.08 ) Reference 1.0 4 (0.3 4 3. 19 ) 5.33 (0. 9 3, 30.50 ) Reference 1.81 (0.47, 6.97) 1.66 (0.40, 6.87) 1.15 (0.22, 6.01) 1.67 (1.11, 2.51) NA 0.69 ND NA ND 0.64 NA ND 0. 3 2 NA 0.49 NA 0.59 NA 0. 59 0.73 NA 0.93 0.05 NA 0.38 0.47 0.86 0.01 OR = crude odds ratios; ** 95% CI = 95% confidence interval; NA = Not applicable; ND = Not determined

PAGE 114

114 Table 3 5 Multivariable analysis of investigated exposure factors at admission associated with Salmonella without diarrhea Variable SE OR 95% CI P Plasma protein Low versus normal High versus normal Triglycerides High versus normal 0.02 1.04 1.21 0.91 0.58 0.54 0.97 2.85 3.37 0.16, 5.88 0. 91, 8.91 1.15, 9.87 0.97 0.07 0.02

PAGE 115

115 Table 3 6 Multivariable analysis of investigated exposure factors (during hospitalization) associated with Salmonella in horses without diarrhea Variable SE OR 95% CI P Model A Abdominal surgery Yes vs. No Number of fecal samples tested Model B Season Summer and Fall vs. Winter & Spring Number of fecal samples tested 1.02 0.53 1.59 0.65 0.59 0.21 0.61 0.23 2.79 1.70 4.92 1.92 0.86, 9.01 1.11, 2. 60 1.46, 16.57 1.20, 3.06 0.08 0.01 0.01 0.005

PAGE 116

116 CHAPTER 4 DIAGNOSTIC PERFORMAN CE OF REAL TIME PCR FOR DIAGNOS IS OF SALMONELLA SHEDDING IN HOSPITALIZED HORS ES Shedding of Salmonella spp in feces of hospitalized horses is an important problem for large animal hospitals because of the potential risk of outbreaks of nosocomial Salmonella infections. Outbreaks of nosocomial Salmonella infections in hospitalized horses can result in temporary hospital closure, loss of revenue, mortality in horses, zoo notic infections, and reduced caseload. 1 4,6,92,128 130,132 To reduce the risk of outbreaks of nosocomial Salmonella infections among hospitalized horses, several veterinary teaching hospitals have established surveillance and infection control programs. I n most veterinary hospitals with a surveillance and infection control program, bacteriological culture is considered the gold standard, 6,8,10,92 and is the most frequently used diagnostic technique for detection of Salmonella spp in fecal specimens of hors es. 6 A major limitation of using bacteriological culture to monitor community acquired and nosocomial Salmonella infections in horses is the time required (3 to 5 days) to obtain laboratory results. In the absence of overt clinical signs (diarrhea or fever and leucopenia), this prolonged detection time creates a delay in implementation of appropriate infection control measures (such as isolation and barrier nursing) which are necessary to minimize the risk of nosocomial infections and environmental contamin ation. Several studies have de scribed PCR based tests for detecting Salmonella spp in fecal specimens in horses. One advantage of using PCR tests is that results can be obtained in 24 hours, thus allowing an earlier implementation of appropriate infection control measures However, despite the availability of rapid and cost effective PCR

PAGE 117

1 17 tests for detection of Salmonella its acceptance as a surveillance tool for monitoring community acquired and nosocomial Salmonella infections in horses has been limited. Two previous studies compared the diagnostic performance of conventional PCR and culture and revealed a number of limitations of PCR as a surveillance tool. 8,11 In one study, the study population included horses with and without clinical signs of salmonel losis and th is conventional PCR assay targeted the transport operon gene of Salmonella 11 In that study, a mong equine outpatients without clinical signs of salmonellosis, 26/152 (17%) tested positive by PCR and 0/152 (0%) tested positive by culture, but on ly one fecal sample was collected from this study population to test for Salmonella The p ossibility exists that the 26 horses that tested positive by PCR were colonized with Salmonella and had the se been tested two or more times by culture, the true statu s of these samples may have been Salmonella culture positive. A gold standard definition for true negatives that considers horses without clinical signs of salmonellosis that are sampled and tested by culture on only one fecal sample likely produces inconc lusive results. B acteriological culture of multiple fecal samples from horses has been shown to yield a greater proportion of Salmonella positive results compared to single fecal samples. 1 40 In a second conventional PCR study that targeted the same gene t he study population included hospitalized horses without signs of GI disease from which multiple fecal samples were collected at 24 hour intervals. 8 R esults revealed a high frequency of Salmonella PCR positive results. In that study, the analysis for speci ficity included 105 horses without clinical signs of salmonellosis that tested Salmonella used as the gold standard, the epidemiologic specificity of the PCR assay used would

PAGE 118

118 be very low, as only 29 horses (27.6%) were classified as PCR negative. In that study many false p ositive results were yielded by that particular protocol. 8 Given the limited bacterial sequencing information available at that time, i t is possible that the primers used in that study were not as specific as presumed and may have cross reacted with other bacteri a Two additional previous studies used real time PCR technology to compare the diagnostic performance of PCR and culture and revealed s imilar limitat ions. In one study that targeted the invA gene, a specificity estimate of 339/345 or 98% was report ed for the PCR assay 13 However, in that study the fecal samples used to estimate specificity were collected from horses and the environment of pens of a horse feedlot. The clinical status of study horses was not reported and it wa s not clear if horses in that study had colic with or without clinical signs of salmonellosis. In addition, only one fecal sample was collected from each study horse. In a second study that targeted the same gene, a relative specificity of 889/905 (98%) was reported for the PCR a ssay 10 T he a nalysis for specificity included 905 horses with and without clinical signs of salmonellosis that tested Salmonella culture negative on one fecal sample under enrichment In these two previous real time PCR studies, 10,13 t he lack of a gold sta ndard definition for true negatives, and the use of horses that tested culture negative on only one fecal sample may have led to m isclassification bias produc ing inconclusive results. Different sampling methods can affect the diagnostic performance assessm ent of diagnostic tests. Wh en assessing the specificity of a PCR assay, i t i s important that a gold standard definition for true negatives be consistently consider ed. In particular one that defines negative horses as those horses without clinical signs of salmonellosis that

PAGE 119

119 are sampled multiple times and test negative by culture In addition it is important that a PCR assay defines what frequency of PCR false positives is acceptable. In a hospital setting, the consequences of false positive results can be considerable because equine in patients with a positive result must be placed in isolation stalls, increasing the cost of hospital fees. For PCR to be accepted as a surveillance tool, the aforementioned limitations in previous conventional and real time PC R studies need to be addressed. The objective of this study was to assess the diagnostic performance of real time PCR assay s for detection of Salmonella spp in fecal specimens of hospitalized horses, compared to bacteriological culture. The limitations obs erved in previous PCR studies were addressed in this study i n an attempt to 1) define in silico the specificity and sensitivity of diagnostic PCR protocols on current bacterial bioinformatical databases, 2) develop a case definition for each population of true positive and true negative Salmonella horses, and 3) develop an algorithm for PCR testing that is reliable, with a high relative specificity, sensitivity, and accuracy. Materials and Methods Study Population The study was conducted at the University o f Florida Large Animal Hospital (UF LAH) The hospital has an on going surveillance program for detection of community acquired and nosocomial Salmonella infections in hospitalized horses. All equine patients with or without clinical signs of gastrointesti nal (GI) disease admitted to the UF LAH during a one year period from January to December 2011 were eligible for inclusion in the study. Horses hospitalized for less than 48 hours or with less than three fecal samples collected for bacteriological culture were excluded. The study protocol

PAGE 120

120 was approved by the Institutional Animal Care and Use Committee at the University of Florida (Protocol # 201104908). Study Design Figure 4 1 is a chart that presents the overall sampling approach that was used to estimate the relative sensitivity and specificity of the real time PCR test for diagnosis of Salmonella in equine feces To assess the diagnostic sensitivity (detection of true positives) of the real time PCR assay, fecal or swab samples from 20 horses with signs of GI disease (colic, diarrhea, or fever and leucopenia) that were classified as Salmonella culture positive were used in the analysis. To assess the diagnostic specificity (detection of true negatives), fecal samples from two different groups of horses we re used in the analysis. The first group included fecal samples from 30 horses without signs of GI disease that were classified as Salmonella culture negative on 5 fecal samples. The second group included fecal or swab samples from 43 h orses with signs of GI disease that were classified as Salmonella culture negative on at least three fecal samples. Bacteriological culture was used as the gold standard test in this study because it is the most frequently used diagnostic method for detection of Salmonella sp p in fecal specimens of horses. 6,8,10 True positive horses were defined as GI equine inpatients that tested Salmonella positive by culture on at least one fecal or swab sample. True negative horses were defined as non GI or GI equine inpatients that had fi ve fecal samples or at least three fecal or swab samples collected respectively, and tested Salmonella culture negative on all samples. The rationale for sampling and testing each study horse at least three times is that the probability of detecting Salmo nella spp in fecal samples increases with the number

PAGE 121

121 of samples tested. Temporal sensitivity is the probability of detecting a pathogen in a given time period and refers to the ability of a diagnostic test to detect the pathogen early, as opposed to late. 1 8 9 Assuming the epidemiologic sensitivity of the PCR assay is 90% (relative to culture), the probability of detecting a horse infected with Salmonella spp upon admission (sample 1) is 0.90, and the probability that the horse would fail to test positive (fa lse negative), would be 1 0.90 = 0.10. If the horse failed to test positive upon admission and is tested again 24 hours later (sample 2), the probability that the horse would then test positive would be 0.10 x 0.90, or 0.09. The cumulative probability t hat the horse would be detected after two samples would be the probability of detection at admission (sample 1) plus 24 h later (sample 2), or 0.90 + 0.09 = 0.99. Finally, if the horse failed to test positive upon admission and 24 h later (samples 1 and 2) the probability that the horse would then test positive 48 h after admission (sample 3) would be 0.10 x 0.10 x 0.90, or 0.009. Thus, the cumulative probability that the horse would be detected after three samples would be the probability of detection at admission (sample 1) plus 24 h after admission (sample 2) plus 48 h after admission, or 0.90 + 0.09 + 0.009 = 0.999. A sample size of 20 horses classified as Salmonella true positive and 30 horses classified as Salmonella true negative would provide 95% co nfidence and 80% power to estimate a desired real time PCR assay sensitivity of 90% (95% confidence interval = 76.8, 100.0) and a specificity of 9 6 % (90.2, 100 ). Fecal Sample Collection For GI horses, fecal sample collection was performed as part of the U F LAH surveillance program for early detection of fecal shedding of Salmonella spp at the time of admission and during hospitalization. A fecal sample or rectal swab (if fecal sample

PAGE 122

122 was not available) was collected from each study horse within 12 h of adm ission (sample 1) and submitted for bacteriological culture. Thereafter, additional fecal samples (samples 2, 3, and more) were collected from the stall floor each morning prior to cleaning at 72 96 hour intervals until the patient was discharged from the hospital. Fecal samples collected outside regular business hours were refrigerated at 4 C prior to laboratory submission (for a maximum of 24 h) For some horses, additional fecal samples were collected at the discretion of the attending clinician (eg, eve ry 12 to 24 hours ). For non GI horses, only fecal samples were collected (ie, no fecal swabs collected) and submitted for bacteriological culture. The first fecal sample (sample 1) was collected within 12 h of admission from the stall floor of each stall housing the study horse by one of the study investigators. Thereafter, additional samples (samples 2 5) were collected at 12 h intervals during hospitalization. As part of bacteriological culture procedures, all fecal or swab samples collected from study horses were enriched in 10 m L of Hajna Tetrathionate broth (with Iodine Iodide) (TTB) for 24 h at 37 C. The following day, the broth was sub cultured on Hektoen Enteric Agar (HEA) plates. The remainder of the broth (approximately 9.5 m L ) was aliquoted and stored at 80 C to be used later for Salmonella DNA extraction and real time PCR testing. Microbiological Procedures Bacteriologic culture of fecal samples for detection of Salmonella spp was performed at the UF Veterinary Clinical Microbiology Laboratory as previously described. 132 Briefly, for selective enrichment, 1 to 2 g of fresh feces was placed in 10 m L of TTB, and incubated at 37 C for 24 hours. The following day, a sample of the TTB

PAGE 123

123 was sub cultured on HEA plates. Plates were incubated at 37 C for 24 hours. Non lactose fermenting, hydrogen sulfide producing colonies were selected and isolated. These colonies were then inoculated on urea agar and lysine iron agar slants and incubated at 37 C for 24 hours. Identification of urease negative and hydroge n sulfide producing organisms was established through use of a commercially available identification system. Serogroup of Salmonella isolates was determined by means of agglutination; polyvalent (A through I and Vi) and group specific (A through E) Salmone lla O antisera were used. Salmonella isolates were tested for antimicrobial susceptibility using the minim al inhibitory concentration (MIC) method with commercially prepared plates and the Kirby Bauer disk diffusion method. Serotyping of Salmonella isolate s was performed at the USDA National Veterinary Services Laboratories in Ames, Iowa. Real time PCR Assay Procedures Bioinformatics analysis T he goal for the bioinformatic analysis was to determine, in silico, the specificity of the primers and probes previ ously investigated for detect i on of Salmonella spp in equine fecal samples Several targets were included consisting of the histidine transport operon, invE, sip, spaQ, and invA gene. 10 13 Primer and probe sequences from published studies were blasted us ing the National Centre for Biotechnology Information ( NCBI ) database to determine t he degree of similarity and homology of Salmonella against bacterial nucleotide sequences. Individual primers and probes were blasted and compared using the following infor mation: number of hits, organisms hit, description (chromosome/plasmid), sequence similarity scores, query coverage, expected values, and percent maxim al identity score. In addition, conventional PCR primers and real time

PAGE 124

124 PCR primers and probes were blaste d as a combination (ie., forward/reverse and forward/probe/reverse). In the end, t wo sets of primers and probes were selected from published real time PCR studies that targeted Salmonella DNA in equine fecal samples. After selection of the two sets of prim ers and probes, the target gene was cut into five overlapping fragments and blasted to verify the existence of sequence alignments with non Salmonella organisms. In addition, the existence of sequence alignments with non Salmonella organisms when using the histidine transport operon gene (Genbank A ccession N umber V01373.1; 3700 bp) was investigated. This investigation was important because this gene had been used in previous studies 8,11 that produced a high proportion of PCR false positive results (26/152 o r 17% 1 1 and 76/105 or 72% 8 ), when bacteriologic culture results were used as the gold standard. Optimization of primer and probe concentration The two selected sets of primers and probes were optimized using the Applied Biosystems Fast 7500 real time PC R system (Applied Biosystems, Carlsbad, CA) and SYBR green and TaqMan probe PCR chemistries. First, each set of primers was evaluated using SYBR green. The rationale for using SYBR green chemistry was to determine if non specific binding occurred (based on dissociation curves). For each set of primers, a primer optimization matrix ( Table 4 1 ) was used to determine the minim al primer concentration that yielded the minim al C T and maxim al Salmonella Typhimurium DNA was used as a template and deionized wat er as a negative control. Samples were amplified using the following thermal cycling protocol with SYBR green: 20 s at 95 C; 3 s at 95 C for 40 cycles, 30 s at 60 C; 15 s at 95 C, 1 min at 60 C, 15 s at 95 C, and 15 s at 60 C. Optimal primer concentrations were determined and

PAGE 125

125 dissociation curve analysis was performed to examine for non specific binding (absence of erroneous peaks/presence of different melting temperature from the specific product). Each set of primers was evaluated using TaqMan probe chemis try to determine the optimal primer/probe concentrations. A primer optimization matrix ( Table 4 1 ) was used to determine the minim al primer concentration that yielded the minim al C T and maxim al probe concentration was varied from 50 to 300 nmols (50/150/200/250/300 nmols) to determine the optimal probe concentration that yielded the minim al C T and maxim al Salmonella Typhimurium DNA was used as a template and deionized water as a negative con trol. Samples were amplified using the following thermal cycling protocol: 20 s at 95 C, 3 s at 95 C for 40 cycles and 30 s at 60 C. A third set of primers and probe that universally target s the 16S rRNA gene of bacteria was used as a control of the effic iency of DNA purification and amplification and as an indicator of fecal inhibition 1 9 0 This third set of primers and probe were optimized following the same approach as previously described. The efficiency of the PCR assay was determined for each set of primers/probes by using seven 10 fold dilutions of Salmonella DNA (10 ng/ L) to generate standard curves. For both sets of primers/probes, the slope of the log linear phase and R 2 values were examined to verify if the amplification of the PCR assay was ef ficient. Analytical specificity and sensitivity of the PCR assay Analytical specificity of the assay (using the two selected sets of primers/probes) and cross reactivity with non Salmonella enteric and non enteric organisms was determined as described prev iously 1 0 with modifications. Enteric and non enteric organisms tested included E. coli, Klebsiella pneumoniae, Bordetella bronchiseptica,

PAGE 126

126 Streptococcus equi subsp z ooepidemicus, Staphylococcus aureus, Clostridium difficile Enterobacter cloacae, Enterobac ter aerogenes, Pasteurella multocida, Pseudomonas aeruginosa, Enterococcus faecalis, Enterococcus casseliflavus, Paeni bacillus, Bacteroides fragilis, Proteus mirabilis, and Streptococcus equi subsp equi. All the non Salmonella strains were obtained from the University of Florida Shands Diagnostic Microbiology laboratory and collections at College of Veterinary Medicine. The analytical sensitivity of the assay (using the two selected sets of primers/probes) was determined using 1 3 Salmonella isolates of eq uine origin from 5 serogroups: serogroup B ( S. Saintpaul and S. Typhimurium); serogroup C1 ( S. Mbandaka, S. Braenderup); serogroup C2 ( S. Newport S. Litchfield, and S. Muenchen); serogroup D ( S. Enteriditis, S Miami, S Javiana) and serogroup E ( S. Anatu m, S Meleagridis, S Muenster). All Salmonella strains were obtained from the UF Veterinary Hospitals Microbiology laboratory. All Salmonella and non Salmonella strains were grown on CBA plates for 24 h at 37 C. A colony from the agar plates was inocula ted in TSB or Thioglycollate with H & K broth ( Bacteroides fragilis ) for 24 h at 37 C. Broth was aliquoted, stored at 20 C and later used for DNA extraction using DNeasy tissue kit (Qiagen Valencia, CA). Real time PCR was performed in duplicates for each isolate. Salmonella Typhimurium DNA was used as a positive control and deionized water as a negative control. Comparison of the analytical sensitivity of DNA extractio n kits Three commercially available DNA extraction kits (DNeasy tissue kit, Qiagen, Valencia, California; UltraClean fecal kit, MOBIO, Carlsbad, California; and QIAamp DNA Stool Mini Kit, Qiagen, Valencia, California) were selected and compared to dete rmine the technique with the optimal analytical sensitivity (lowest C T or highest

PAGE 127

127 number of Salmonella gene copies/u L DNA). The DNeasy tissue kit was selected for evaluation because it was reported in a previous study to have optimal sensitivity (compared to other kits) during extraction of Salmonella DNA from equine feces. 12 The UltraClean fecal kit and the QIAamp DNA Stool Mini Kit were selected for evaluation because these kits are manufactured specifically for DNA extraction from fecal samples. The D NA extraction kit with the highest an alytical sensitivity (lowest C T or highest number of Salmonella gene copies/u L DNA) was used to extract DNA from all study samples. A Salmonella Typhimurium isolate obtained from the UF Veterinary Hospitals Microbiolog y laboratory was streaked on CBA plates and incubated for 24 h at 37 C. The following day, one colony from the CBA plate was inoculated in 5 m L of TSB and incubated for 24 h at 37 C. The following day, ten fold dilutions of the 5 ml TSB were performed in p hosphate b uffered s aline (PBS). An aliquot of 50 u L of each dilution was spread on HEA plates using the standard spread plate technique and the agar plates were incubated at 37 C for 24 h. The following day, Salmonella colony forming units (CFU) on each pl ate were enumerated and the dilution with approximately 10 9 CFU/m L was determined. Using a predetermined number of CFU (10 9 CFU/m L ), 1 ml of the selected dilution was used to spike 1 g of horse feces. The spiked horse feces were collected from a clinically normal horse. A horse was classified as clinically normal based on physical examination findings and absence of signs of GI disease, diarrhea, fever, or leucopenia, and after testing Salmonella negative by bacteriological culture and real time PCR on five consecutive fecal samples collected at 24 h intervals. Each spiked fecal sample was diluted in 10 ml of TTB and incubated at 37 C for 24h.

PAGE 128

128 Thereafter, the 10 m L of TTB was aliquoted into nine samples (1 m L each) and stored for DNA extraction using the thr ee different kits ( three samples per kit). Three trials of the above experiment were performed. Real time PCR was performed on the DNA samples and C T (number of cycles required for the fluorescent signal to cross the threshold) for the 3 different extracti on kits were compared for each trial and with all trials combined to determine the kit with the lowest C T The kit with the lowest C T values was selected and used to extract DNA from all the study samples. Evaluation of PCR assay detection limit The detect ion limit of the PCR assay was evaluated using Salmonella organisms in liquid media ( PBS ) and fecal samples. The PCR assay detection limit using Salmonella organisms suspended in PBS was determined as described previously 10,14 3 ,14 4 with modifications. A S almonella Typhimurium isolate ( UF Veterinary Hospitals Microbiology laboratory Gainesville, FL) was streaked on CBA plates and incubated for 24 h at 37 C. The following day, one colony from the CBA plate was inoculated in 5 m L of TSB and incubated for 24 h at 37 C. This was then diluted 10 fold in PBS. An aliquot of 50 u L of each dilution was spread on HEA plates using the standard spread plate technique and the agar plates were incubated at 37 C for 24 h. The following day, Salmonella colony forming units on each HEA plate were enumerated twice and the dilution with approximately 10 9 CFU/m L was determined. Using the selected dilution with 10 9 CFU/m L a ten fold dilution was performed with PBS. The dilutions were aliquoted, stored at 20 C, and later DNA wa s extracted using the selected optimal DNA extraction kit. Three trials of this experiment were performed. Real time PCR was performed in duplicates for each DNA sample using both sets of

PAGE 129

129 selected primers and probes. Salmonella Typhimurium DNA was used as a positive control and deionized water as a negative control. The PCR assay detection limit using spiked fecal samples was determined as described previously with modifications. 14 3 ,14 4 Enumeration of Salmonella colony forming units ( 10 9 CFU/m L ), was perfor med as described under evaluation of PCR assay detection limit using PBS (above). Using the selected dilution (with 10 9 CFU/m L ), a ten fold dilution was performed with PBS. One m L of each dilution was used to spike 1 g of horse feces. The spiked horse fece s were collected from a clinically normal horse. A horse was classified as clinically normal based on physical examination findings and absence of signs of GI disease, diarrhea, fever, or leucopenia, and after testing Salmonella negative by bacteriological culture and real time PCR on five consecutive fecal samples collected at 24 h intervals. Each spiked fecal sample was diluted in 10 m L of TTB and incubated at 37 C for 24h. The following day, the TTB was aliquoted, stored at 20 C, and later DNA was extra cted using the selected optimal DNA extraction kit. Three trials of this experiment were performed. Real time PCR was performed in duplicates for each DNA sample using both sets of selected primers and probes. Salmonella Typhimurium DNA was used as a posit ive control and deionized water as a negative control. Detection of Salmonella DNA in study samples To minimize observation bias each broth sample was independently masked before DNA extraction and real time PC R testing. After PCR testing, results where unmasked before data analysis. Each s tudy sample w as tested in duplicates using the two selected primer sets. Results were reported qualitatively (positive or negative). A threshold cycle value of 37

PAGE 130

130 was considered a positive result for each of the two sel ected primer sets based on the standard curve for each primer set where the lowest concentration of template was detected at a threshold cycle value of 37 ( 1x10e 5 ng/ul ; Salmonella DNA ) Data Collection For each study horse, the following data were colle cted: medical record number, admission date, discharge date, duration of hospitalization, age, gender, breed, presenting complaint, clinical findings at admission and during hospitalization (colic: yes, no; diarrhea: yes, no; fever: yes, no; leucopenia: ye s, no), clinical procedures during hospitalization (treatment with antimicrobials: yes, no; surgery: yes, no), diagnosis at time of discharge, type of sample collected (fecal sample or rectal swab), fecal sample identification number (i.e., 1,2,3), date fe cal sample was collected, number of fecal samples collected, culture results ( Salmonella positive or negative), and PCR results ( Salmonella positive or negative). For horses that tested positive by bacteriological culture, information on Salmonella isolate s was collected including serogroup, serotype, fecal sample number that first tested positive, hospital day when first test positive sample was collected, and number of hospital days when first test positive sample was collected. Data Analysis The W ilcoxo n S igned R ank test was used to compare mean C T values for the UltraClean fecal kit and QIAamp DNA Stool Mini Kit to determine the kit with the lowest C T These data were analyzed as paired data at each of the 3 trial s performed (each trial on a separate plate) because efficiency of a PCR reaction and therefore C T values can vary from plate to plate.

PAGE 131

131 Among GI inpatients, the epidemiologic sensitivity of the PCR assay s w ere calculated by dividing the number of horses that tested positive to PCR on 1 or mo re samples by the total number of horses classified as true positives. Among, GI and non GI inpatients, the epidemiologic specificity was calculated by dividing the number of horses that tested negative to PCR on all samples by the total number of horses c lassified as true negatives. The positive predictive value of the PCR assays were calculated by dividing the number of horses that tested positive to both PCR and culture on 1 or more samples by the total number of horses that tested positive to PCR. The negative predictive value of the PCR assays were calculated by dividing the number of horses that tested negative to both PCR and culture on 1 or more samples by the total number of horses that tested negative to PCR. Results Bioinformatics Analysis A summ ary of the blast results for i ndividual primers and probes are shown in Table 4 2 Two sets of primers and probes were selected from published PCR studies performed in horses. The criteria for selection involved blasting the published primers and probes to determine the degree of similarity and homology of Salmonella against bacterial nucleotide sequences. In addition, published conventional PCR primers and real time PCR primers and probes were blasted as a combination (ie., forward/reverse and forward/pro be/reverse). When primers for the histidine transport gene (Cohen et al 1996) 1 1 were blasted as forward/reverse, the number of Salmonella hits/total number of organism hits was 33/264 or12%. When each set of real time primers and probes were blasted as

PAGE 132

132 fo rward/probe/reverse, the number of Salmonella hits/total number of organism hits was highest for invA (78/100 or 78%; Pusterla et al 2009 1 0 ) followed by sip (24/41 or 59%; Kurowski et al 2002 1 2 ), invA (67/152 or 44%; Bohaychuk et al 2007 13 ), spaQ (47/14 5 or 32%; Kurowski et al 2002 1 2 ), and invE (2/12 or 16% ; Kurowski et al 2002 1 2 ). Overall, the two sets of primers and probes that had a combination of maximal identity with Salmonella spp and least identity with non Salmonella organisms were those publ ished by Pusterla et al 2009 1 0 and Bohaychuk et al 2007. 13 Nucleotide sequences of the two selected sets of primers and probes and the universal bacteria primer used as a control are shown in Table 4 3 The selected probes target the invasion A (invA) g ene of Salmonella. The invA gene is located on p athogenecity i sland 1 of the Salmonella chromosome and is essential for epithelial invasion, and is present in all invasive strains of Salmonella and absent from closely related organisms such as E. coli. 19 1 1 9 3 To verify the existence of sequence alignments with non Salmonella organisms, the invA gene (Genbank accession number U43271; 1950 bp) was cut into five overlapping fragments and blasted. Results are presented in Table 4 4 The invA gene had sequence a lignment with one non Salmonella organism ( Photobacterium profundum ; a Gram negative bacteria found mainly in marine environments). The existence of sequence alignments with non Salmonella organisms for the histidine transport operon gene (Genbank accessio n number V01373.1; 3700 bp) was also investigated. This investigation was performed because this gene had been used in previous studies that produced a high proportion of PCR false positive results (26/152 or 17% and 76/105 or 72%, respectively), when bact eriologic culture results were used as the gold standard. 8,11 A summary of the blast results are shown in Table 4 4 Results

PAGE 133

133 from this investigation revealed that the histidine transport operon gene had sequence alignments with several non Salmonella org anisms (including Escherichia coli, Enterobacter cloacae, Shigella sonnei Shigella flexneri, Shigella boydii, Escherichia fergusonii, Citrobacter koseri, Citrobacter rodentium, Cronobacter turicensis Serratia sp., Burkholderia cenocepacia Enterobacter a sburiae, Dickeya dadantii, Pantoea ananatis, Pectobacterium wasabiae, Pseudomonas syringae, Pseudomonas brassicacearum, Pantoea vagans, and Erwinia pyrifoliae ). Optimization o f Primer a nd Probe Concentration First, each set of selected primers was evaluat ed using SYBR green. For each set of primers, a primer optimization matrix was used to determine the minim al primer concentration that yielded the minim al C T and maxim al f orward and r everse) concentrations of 300, 100, and 30 nmols for primers 1, 2 and 3 respectively were optimal. For each of the three primer sets, non specific binding was not observed at the different primer concentrations. Figures 4 2 4 3 and 4 4 show dissociation curves for primer sets 1 3. Each set of primers and probes wa s evaluated using TaqMan probe chemistry. A primer optimization matrix was used to determine the minimum primer concentration that yielded the minimum C T and maxim concentration, varying probe concentrations were used to determine the optimal probe concentration (that yielded the minimum C T and (f orward and r everse) concentrations for primers 1, 2, and 3 were 300, 100, and 30 nmols respectively. Optimal probe concentrations for probe 1, 2, and 3 were 250, 100, and 250 nmols respectively.

PAGE 134

134 The efficiency of the PCR assay was determined for each set of primers/probes by using seven logs of 10 fol d dilutions of Salmonella DNA (10 ng/ L ) to generate standard curves. For each set of primers/probes, the slope of the log linear phase and R 2 values showed that amplification of the PCR assay was efficient, as shown in figure 4 5 (Primer set 1: Slope = 3 .55, Intercept = 18.12 R 2 = 0.99 ) and figure 4 6 ( primer set 2: Slope = 3.56, Intercept = 19.38, R 2 = 0.99 ) Analytical Specificity a nd Sensitivity The two selected sets of primers/probes were specific for Salmonella ; there was no cross reactivity with any of the enteric and non enteric non Salmonella organisms that were tested. All 13 Salmonella isolates tested positive using the two selected sets of primers/probes. Comparison o f t he Analytical Sensitivity o f DNA Extraction Kits For primer sets 1 and 2 mean C T values for the three DNA extraction kits (DNeasy tissue kit, UltraClean fecal kit, and QIAamp DNA Stool Mini Kit) were compared for each of the 3 trials that were performed in triplicates to determine the kit with the lowest C T (Tables 4 5 an d 4 6). The UltraClean fecal kit had significantly lower mean C T values compared to the QIAamp DNA Stool Mini Kit using primer set 1 ( P < 0.009) or primer set 2 ( P < 0.009). No C T results were reported for t he DNeasy tissue kit, suggesting that this kit could not extract Salmonella DNA from horse fecal samples that were spiked and enriched. W hen this experiment was re run using samples in which PBS was spiked with S Typhimurium and horse fecal samples that were spiked with S Typhimurium and enriched, C T values were reported for spiked PBS and still no C T values were reported for spiked feca l samples. Results of the third primer set (universal bacteria primer) used as a control of the efficiency of DNA purification

PAGE 135

135 showed that no bacterial DNA was extrac ted from spiked horse fecal samples that were enriched but bacterial DNA was extracted from spiked PBS. On the basis of the results of the three DNA kits, t he UltraClean fecal kit had the highest sensitivity (lowest mean C T values ) and was used to extract DNA from all the study samples. DNeasy tissue kit was not used further for extraction of Salmonella DNA from horse fecal samples that were spiked and enriched. PCR Assay Detection Limit The assay detect ed as low as < 1 colony forming unit when Salmonella DNA was purified from PBS using the U ltraClean fecal DNA extraction kit. The assay detect ed as low as < 1 colony forming unit when Salmonella DNA was purified from spiked fecal samples using the UltraClean fecal DNA extraction kit. Detection of Salmonella DNA in Study Samples Study fecal s a mples were collected from 93 horses over a 12 month period and the 93 horses included 20 horses that presented with GI signs and tested positive for Salmonella on at least 1 fecal or swab sample, 43 horses that presented with GI signs and tested negative f or Salmonella on at least 3 fecal or swab samples, and 30 horses that presented without GI signs and tested negative for Salmonella on 5 fecal samples. Horse breeds sampled included the following: Quarter Horses (19); Thoroughbred (16); Cross (9); Warm bl ood (8); Percheron (5); American Paint Horse (4); Arabian (3); Hanoverian (3); Pony (3); Appaloosa (3), Tennessee Walking Horse (3); and Others (17). Median age of study horses was 9 yrs (1 st quartile = 3.5 3 rd quartile = 13 ; range = 33 ). Thirty five of the 93 sampled horses were females, 46 were castrated males and 12 were non castrated males.

PAGE 136

136 The median number of samples collected from horses that presented with GI signs and tested positive was 3 (1 st quartile = 2 3 rd quart ile = 4.5 ), for horses that presented with GI signs and tested negative was 4 (1 st quartile = 3 3 rd quartile = 5 ), and for horses that presented without GI signs and tested negative was 5 (1 st quartile = 5 3 rd quartile = 5 ). H orses that were classified a s true positives (with GI signs and tested positive for Salmonella n=20) presented for colic (18/20), diarrhea (1), and fever (1). Horses that were classified as true negatives (without GI signs and tested negative, n=30) presented with lameness problems (8), o cular problems (16 ; melting corneal ulcer (7), ocular discharge (1), stromal abscess (1), uveitis (2), conjunctivitis, tearing/swelling/blepharospasm (5) ) and other problems ( 6; septic carpus (1), canker/white line disease (1), splint bone removal ( 1), foreign body in hind foot (1), inguinal hernia surgery (1), hind limb abscess (1)). Relative S pecificity Non GI horses. The relative specificity of the assay usi ng primer set 1 for samples # 1 to 5 varied from 26/30 (87%) to 30/30 (100%) (T able 4 7 ). Overall, 7 / 150 samples were classified as fa lse positive on real time PCR compared to bacteriological culture. The median threshold cycle ( C T ) values for the se 7 samples was 34.70 (1 st quartile = 33.28, 3 rd quartile = 35.29; rang e = 31.98 to 37.00). Relat ive specificity of the assay using primer set 2 for samples # 1 to 5 varied from 29/30 (97%) to 30/30 (100%). Overall, 5 / 150 samples were classified as false p ositive on real time PCR compared to bacteriological culture. The median C T for the se 5 samples was 33.19 (32.34, 34.48; 32.33 to 35.31).

PAGE 137

137 The relative specificity for the real time PCR assay when using both primer sets 1 and 2 for samples # 1 to 5 varied from 26/30 (87%) to 30/30 (100%) GI horses. The relative specificity of the assay using primer set 1 for samples # 1 to 3 varied from 36/41 (88%) to 42/43 (98%) (Table 4 8). Overall, 10 / 129 samples were classified as false positive on real time PCR, compared to bacteriological culture. The median C T values for these 10 samples = 35.39 (1st quartile = 26.93, 3rd quartile = 36.00; range = 22.03, 36.95). Relative specificity of the assay using primer set 2 for samples # 1 to 3 varied from 36/42 (86%) to 41/43 (95%). Overall, 12 / 129 samples were classified as false positive on real time PCR, compared to bacteriological culture. The median C T for these 12 samples was 34.45 (1st quartile = 30.57, 3rd quartile = 36.09; range = 21.44, 36.70). The relative specificity for the real time PCR assay when using both primer sets 1 and 2 for samples # 1 to 3 varied from 36 / 4 3 (8 4 %) to 41 / 4 3 ( 95 %) Relative S ensitivity Relative sensitivity of the assay using primer set s 1 and 2 for samples # 1 to 7 = 20/20 (100%). Eighteen of 20 GI horses tested positive for the first time both by PCR and culture on the same sample n umber (Table 4 9 ) Two of 20 GI horses tested posi tive by PCR earlier compared to culture ( Table 4 9 ; horse No. 11 and horse No. 20) Positive and Negative Predictive Values Non GI horses. For primer set 1, on the basis of a specificity of 28/30 ( 93%) fo r sample # 1 and sensitivity of 20/20 ( 100% ) the positive and negative predictive values were estimated to be 20/22 ( 91%) and 28/28 ( 100%) respectively. For primer set 2, based on a specificity of 29/30 ( 97%) for sample # 1 and sensitivity of 20/20 ( 100%) the

PAGE 138

138 positive and negative predictive values were estimated to be 20/21 ( 95%) and 29/29 ( 100%) respectively. GI horses. For primer set 1, based on specificity of 42/43 ( 98%) for sample # 1 and sensitivity of 20/20 ( 100%), the positive and negative predict ive values were estimated to be 20/21 ( 95%) and 42/42 ( 100%) respectively. For primer set 2, based on specificity of 41/43 ( 95%) for sample # 1 and sensitivity of 20/20 ( 100%), the positive and negative predictive values were estimated to be 20/22 ( 91%) an d 41/41 ( 100%) respectively. Discussion We have optimized and evaluated the diagnostic performance of a real time PCR assay for detection of Salmonella spp in fecal specimens of hospitalized horses The data presented also includes the side by side primer and probe optimization with both S YBR Green and TaqMan probe technology to confirm that no background fl uor escence was generated that could produce false positive results. Furthermore a bio informatical analysis is presented that was performed against up t o date bacterial sequence databases in order to assess assay specificity in silico. The assay includes enrichment of fecal samples in tetrathionate broth for 12 24 hours, extraction of DNA using a commercially available DNA kit ( UltraClean fecal kit ), and running of the real time PCR assay. The two primer sets used in this PCR assay target the invasion A (invA) gene of Salmonella. The invA gene is located on Pathogenecity Island 1 of the Salmonella chromosome and is essential for epithelial invasion, and is present in all invasive strains of Salmonella and absent from closely related organisms such as E. coli. 1 9 1 1 9 3

PAGE 139

139 In the present study, all fecal or swab samples were enriched. E nrichment of fecal specimen prior to PCR amplification has been shown to impro ve sensitivity of a PCR assay for diagnosis of Salmonella spp in equine feces. 14 3 In a previous conventional PCR study in which enrichment was used prior to PCR amplification, PCR detected 10 0 CFU of S. Enteritidis/g of feces, 14 3 and in another study, perf ormed by the same author, where enrichment was not used, PCR detected 10 3 10 4 CFU of Salmonella spp/g of feces. 7 In another study using real time PCR to detect Salmonella spp in 911 fecal samples without enrichment and enriched (broth) samples, three of t he 911 (0.3%) fecal samples without enrichment tested positive while 22 (2.4%) enriched broth samples tested positive. 10 In the present study, all fecal or swab samples were enriched in tetrathionate broth for 24 h at 37C as part of bacteriological cultur e procedures The following day, part of the broth was sub cultured on Hektoen Enteric Agar plates (for culture) an d the remainder of the broth was aliquoted and stored at 80C and later used for Salmonella DNA extraction and real time PCR testing. In thi s study, the UltraClean fecal kit had the highest sensitivity (lowest mean C T values) compared to the QIAamp DNA Stool Mini Kit and DNeasy tissue kit when used to extract Salmonella DNA from spiked horse fecal samples that were enriched in tetrathionate broth. No C T values were reported for t he DNeasy tissue kit indicating that this method of DNA extraction was not appropriate for extract ing Salmonella DNA from horse fecal samples that were spiked and enriched. However when this experiment was re run us ing samples in which PBS was spiked with S Typhimurium or horse fecal samples that were spiked with S Typhimurium and enriched, C T values were reported for spiked PBS but again no C T values were reported for spiked and

PAGE 140

140 enriched fecal samples. The se resul ts suggest that horse fecal samples that are enriched may contain inhibitors of PCR which could not be removed using the DNeasy tissue kit. This suggestion was supported by results of the third primer set (universal bacteria primer) that was used as a control of the efficiency of DNA purification and as an indicator of fecal inhibition. No C T values were reported using the third primer set. W hen this experiment was re run using samples in which PBS was spiked with S Typhimurium or horse fecal samples that were spiked with S Typhimurium and enriched, C T values were reported for spiked PBS but again no C T values were reported for spiked and enriched fecal samples. The finding in the present study that the DNeasy tissue kit could not extract Salmonella DNA from spiked horse fecal samples that were enriched differs from the finding in a previous real time PCR study 12 In that study, the relative sensitivity and specificity of a PCR assay when using three different commercially available DNA extraction kits to extract Salmonella DNA from enrichment broth was compared D ifferent sensitivities and specificities for the PCR assay were reported for the 3 kits : 100 and 98%; 80 and 100%; and 93 and 100% and t he DNeasy t issue kit had a sensitivity and specificity of 100 and 98% and was considered optimal 12 In th e present study, a bioinformatics analysis was performed to investigate if the target gene, invA was specific for Salmonella enterica In addition, the histidin e transport operon gene targeted in previous studies 8,11 that produced a high proportion of PCR false positive results (26/152 or 17% and 76/105 or 72%, respectively) was investigated. Results showed that the invA gene had sequence alignment with one non S almonella organism only; Photobacterium profundum This organism is a Gram

PAGE 141

141 negative bacterium found mainly in marine environments and was not considered biologically relevant to horses. However, results for the histidine transport operon gene revealed that this gene had sequence alignments with several non Salmonella organisms including Escherichia coli, Enterobacter spp, Shigella spp, and Pseudomonas spp. Some of t hese non Salmonella organisms are biologically relevant because they can be found in horses. These results may explain the high proportion of PCR false positive results identified in previous studies 8,11 In th e present study, the o verall relative specificity of the real time PCR assay when using primer set 1 for non GI and GI negative horses was 87 to100% and 88 to 98% respectively compared to bacteriologic culture A previous real time PCR study 10 that targeted the invA gene and used the same primer set reported a specificity of 904 / 9 05 or 9 9 % when fecal samples with out enrichment were used and 889/905 or 98% when fecal samples with enrichment were used. Although the results for specificity when using enriched fecal samples appear comparable, there are major differences in both studies. In our study, enriched fecal samples (only) were tested by c ulture and PCR. In addition, a total of 5 fecal samples collected at 24 hour intervals f rom non GI equine inpatients were examined In contrast, in the previous study 10 the 905 horses used to assess specificity included horses with or without colic and o nly one fecal sample was tested from each horses I t is possible that some of the se horses with colic were infected with Salmonella and could not be detected based on one culture result only. The lack of a gold standard definition for true negatives, and t he use of horses that tested culture negative on one fecal sample only can lead to misclassification bias which produces inconclusive results.

PAGE 142

142 Overall relative specificity of the assay when using primer set 2 for non GI and GI negative horses was 97 to100% and 86 to 95% respectively A previous real time PCR study that targeted the invA gene and used the same primer set reported a comparable specificity estimate of 339/345 or 98 %. 13 However in that study, fecal samples used to estimate specificity were col lected from horses and the environment of pens of a horse feedlot. The clinical status of study horses was not reported; therefore it is not clear if horses in that study had colic with or without clinical signs of salmonellosis. In addition, in that study only one fecal sample was collected from each study horse. The lack of a gold standard definition for true negatives, and the use of horses that tested culture negative on one fecal sample only can lead to misclassification bias which produces inconclusi ve results. In the present study, the se two potential limitations were addressed by using multiple f ecal samples collected from horses classified as true negatives ( defined as horses without signs of GI disease (including colic, diarrhea, fever, or leucope nia ) that tested Salmonella negative on 5 fecal samples by bacteriological culture ) For non GI horses, a total of five samples were classified as PCR false positive results using primer set 1 and primer set 2, and two additional samples were classified a s false positive by primer set 1. For GI negative horses, a total of ten samples were classified as PCR false positive results using primer set 1 and primer set 2, and two additional samples were classified as false positive by primer set 2. T he C T values for the se PCR false positive s amples were relatively high when compared to the cut off C T value ( 37) used in the present study The relatively high C T values for these samples may be explained by presence of small quantities of initial template DNA ( Salm onella ) in

PAGE 143

143 these samples. F alse positive results can represent replication of very small in i tial numbers of Salmonella DNA in broth samples. 10 In this study, positive and negative predictive values for the PCR assay were estimated for primer set 1 and 2 For primer set 1, t he positive and negative predictive values when using non GI horses were 91 and 100% respectively and t he positive and negative predictive values when using GI negative horses were 95 and 100 % respectively. For primer set 2, the positiv e and negative predictive values when using non GI horses were 95 and 100% respectively and the positive and negative predictive values when using GI negative horses were 91 and 100% respectively. From a clinician and hospital surveillance perspective, a PCR assay with a negative predictive value of 100% is important. This value i ndicate s that the probability that a horse is not shedding Salmonella following a PCR negative test is high or equal to one These results suggest that th e PCR assay evaluated in the present study is reliable in ruling out horses that are not shedding Salmonella and can be used as a diagnostic tool in a Salmonella surveillance pr ogram. The overall relative sensitivity of the real time PCR assay when using primer set 1 was 100%. Th e high sensitivity estimate when using primer set 1 is likely explained by the fact that culture positive samples from GI horses classified as true positive were used to assess sensitivity. In addition, results of analytical sensitivity which included 13 S almonella isolates from five different serogroups were all positive using primer set 1. In a previous real time PCR study 10 that targeted the invA gene and used the same primer set, a sensitivity of 100% was reported although the assessment was limited to 6

PAGE 144

144 horses only that were classified as Salmonella culture positive, and 4 of 6 horses had signs of GI disease. The overall relative sensitivity of the real time PCR assay when using primer set 2 was 100%. The high sensitivity estimate when using primer se t 2 is likely explained by the fact that culture positive samples from GI horses classified as true positive were used to assess sensitivity. In addition, results of analytical sensitivity which included 13 Salmonella isolates from five different serogroup s were all positive using primer set 2. Th e sensitivity estimate in the present study was comparable to that reported i n a previous real time PCR study 13 that targeted the invA gene and used the same primer set ; sensitivity was 28/28 or 100%. In that study 13 however, the clinical status (GI or non GI) of horses used in the evaluation was not reported This study had several strengths and a few limitations. The main strength of this study was the use of standard definitions for horses classified as true ne gatives and true positives to assess specificity and sensitivity respectively By using horses classified as true negatives and true positives, this study addressed the sampling limitations observed in previous PCR studies. In addition, t he present study u sed two different groups of horses to assess specificity; non GI horses that tested culture negative on five samples and GI horses that tested culture negative on three samples. The use of GI horses that tested negative on three culture samples allowed the evaluation of the diagnostic performance of the PCR assay using a representative population that is typically targeted in most hospital Salmonella surveillance program s Therefore our study findings can be extrapolated to hospitalized horses with and with out signs of GI disease. A potential limitation of this study is reported specificity and

PAGE 145

145 sensitivity estimates cannot be extrapolated to other hospital settings because of differences in study population s and PCR technology ( type of PCR machines and mater ials ). On the basis of the relative specificity and sensitivity estimates and positive and negative predictive values determined using horses classified as true negatives and true positives we have evaluated and optimized a real time PCR protocol using either primer set 1 or 2 that is reliable, with a relatively high specific ity and sensitiv ity compared to bacteriological culture. This real time PCR protocol can be used as a surveillance tool to detect Salmonella spp in fecal specimen s of hospitalized h orses. Considering its advantage of being rapid (reporting laboratory results approximately 48 hours later ), compared to culture, this PCR protocol can be used as a screening test that may be confirmed by culture if a samples tests PCR positive. In additio n, culture would provide additional data (a ntibiogram and serotyping data ) that are needed to establish if there is an epidemiological relationship between Salmonella cases or isolates, thus providing evidence that Salmonella infection or colonization is n osocomial in origin.

PAGE 146

146 Table 4 1. Primer optimization matrix used to determine the minimum primer concentration that yielded the minimum C T Forward Primer (nM) Reverse Primer (nM) 50 300 900 50 50/50 50/300 50/900 300 300/50 300/300 300/900 900 900/50 900/300 900/900

PAGE 147

147 Table 4 2 A summary of nucleotide blast results of published individual primers and probes Target G ene Query Template Total number of hits Total number of organism s Number of Salmonella hits Salmonella % max identity (score range) Number of n on Salmonella hits Non Salmonella % max identity (score range) Reference number histidine transport operon F wd 198 1 96 32 95 100 164 94 100 11 Rev 241 232 30 100 202 90 100 invE Fwd 256 251 2 100 249 93 100 12 Rev 622 622 31 100 591 100 Probe 431 362 54 100 308 100 Sip Fwd 835 827 27 100 800 100 12 Rev 457 448 26 100 422 94 100 Probe 659 659 25 90 100 63 4 94 100 spaQ Fwd 382 382 51 95 100 329 94 100 12 Rev 270 270 49 100 206 93 100 Probe 483 483 51 100 430 94 100 invA Fwd 227 227 74 100 149 94 100 13 Rev 643 643 61 100 571 94 100 Probe 314 314 80 95 100 231 94 100 invA Fwd 146 146 59 96 10 0 16 91 100 10 Rev 390 390 66 88 100 324 94 100 Probe 171 124 71 90 100 53 94 100

PAGE 148

148 Table 4 3 Nucleotide sequences of the two selected sets of primers and probes and universal bacteria primer and probe Target gene Reagent Sequence Reference # invA Fwd primer 1 CATTTCTATGTTCGTCATTCCATTACC 10 Rev primer 1 AGGAAACGTTGAAAAACTGAGGATTCT Probe 1 TCTGGTTGATTTCCTGATCGCACTGAATATC invA Fwd primer 2 AACTTCATCGCACCGTCA 13 Rev primer 2 TATTGTCACCGTGGTCCAG Probe 2 TCGGCATCAATACTCATCTGTTTACCG Univ B act Fwd primer GGATGATCAGCCACACTGGA 1 9 0 Rev primer CCAATATTCCTCACTGCTGCC Probe CCCGTAGGAGTCTGGACCGTGTCTCA

PAGE 149

149 Table 4 4 A summary of nucleotide blast results for the invA and histidine transport operon genes Target Gene Query Template (bp) Total number of hits Total number of organism hits Number of Salmonell a hits Salmonella % max identity (score range) Number of n on Salmonella hits Non Salmonella % max identity (score range) invA (1950bp) 1 452 70 70 70 94 100 0 None 402 853 71 71 70 92 100 1 93 803 1254 66 66 66 90 100 0 None 1204 1655 67 67 67 91 100 0 None 1605 1950 60 60 60 92 100 0 None histidine operon (3700bp) 1 841 40 40 31 91 100 9 77 85 741 1581 33 33 28 90 100 5 81 88 1481 2321 33 33 26 91 100 7 72 95 2221 3061 100 74 26 94 100 48 73 85 2961 3700 31 31 27 91 100 4 76 79

PAGE 150

150 Table 4 5. Mean C T values for the DNeasy tissue kit, UltraClean fecal kit, and QIAamp DNA Stool Mini Kit for 3 trials performed in triplicates using Primer set 1. Trial DNeasy Ki t Mean C T UltraClean Kit Mean C T QIAamp Kit Mean C T 1 17.17 19.13 1 17.59 18.81 1 16.70 19.27 2 16.25 18.80 2 16.61 18.76 2 16.07 18.51 3 15.88 19.43 3 16.12 19.51 3 16.09 19.12 Table 4 6. Mean C T values for the DNeasy tissue kit, UltraClean fecal kit, and QIAamp DNA Stool Mini Kit for 3 trials performed in triplicates using Primer set 2. Trial DNeasy Kit Mean C T UltraClean Kit Mean C T QIAamp Kit Mean C T 1 16.80 19.56 1 17.29 19.02 1 16.70 19.13 2 15.85 18.32 2 16.11 19 .02 2 15.64 18.26 3 15.48 19.16 3 15.88 19.74 3 15.68 19.29

PAGE 151

151 Table 4 7 Relative specificity of the real time PCR assay determined using fecal samples from non GI horses that tested culture negative on five consecutive samples Fecal sample numbe r Specificity (%) Primer set 1 Specificity (%) Primer set 2 1 28/30 (93) 29/30 (97) 2 26/30 (87) 27/30 (90) 3 30/30 (100) 30/30 (100) 4 29/30 (97) 29/30 (97) 5 30/30 (100) 30/30 (100) Table 4 8. Relative specificity of the real time PCR assay dete rmined using fecal samples from GI horses that tested culture negative on atleast three samples Fecal sample number Specificity (%) Primer set 1 Specificity (%) Primer set 2 1 42/43 (9 8 ) 41/43 (9 5 ) 2 38 / 42 ( 90 ) 36 / 42 ( 86 ) 3 3 6 / 41 ( 88 ) 3 7 / 41 ( 90 ) Ta ble 4 9 Bacteriological culture and r eal time PCR results for fecal samples collected from GI horses that tested Salmonella positive and were classified as true positives (n= 20 ) Horse Samp #1 Samp # 2 Samp # 3 Samp # 4 Samp # 5 Samp # 6 Samp # 7 1 + 2 + + 3 + + 4 + + + 5 + + 6 + 7 + + + 8 + 9 + + 10 + 11 + + 12 + 13 + 14 + 15 + 16 + 17 + 18 + 19 + 20 + Samples that tested positive by real time PCR using primer set 1 and 2 are highlighted in gray. Horse number 11 and 20 tested culture negative and PCR positive on sample number 1.

PAGE 152

152 Figure 4 1. Sampling approach used to estimate the relative sensi tivity and specificity of the real time PCR assay

PAGE 153

153 Figure 4 2 Dissociation curve analysis for primer set 1 using SYBR Green : showing one melting peak for the primer concentrations that were evaluated i ndicat ing absence of non specific primer bindin g.

PAGE 154

154 Figure 4 3 Dissociation curve analysis for primer set 2 using SYBR Green : showing one melting peak for the primer concentrations that were evaluated i ndicat ing absence of non specific primer binding

PAGE 155

155 Figure 4 4 Dissociation curve analysis for t he universal bacteria primer set using SYBR Green : showing one melting peak for the primer concentrations that were evaluated i ndicat ing absence of non specific primer binding

PAGE 156

156 Figure 4 5 Standard curve for primer set 1

PAGE 157

157 Figure 4 6 Standard curve for primer set 2

PAGE 158

158 CHAPTER 5 AWARENESS AND RELEVA NCE OF HOSPITAL SURV EILLANCE AND INFECTI ON CONTROL SERVICES AMO NG REFERRAL VETERINA RIANS AND CLIENTS At a referral hospital referring veterinarians (RDVMs) and clients/animal owners play a vital role in a hospital surveillance and infection control program Veterinarians participate in the program by referring clients to a tertiary care hospital because of its specialized expertise and need for specialized care. In turn, clients bring high risk patients suc h as those with signs of gastrointestinal (GI) disease that may be targeted in a hospital surveillance and infection control program. At some hospitals, c lients pay for costs related to surveillance and infection control procedures. At the UF LAH, horses t hat present with signs of GI disease are hospitalized for 3 4 days on average and 2 fecal samples are collected and submitted for bacteriological culture testing, resulting in a cost of approximately $56 ($28 per sample) that is paid by the client At a r eferral hospital, information about hospital survei llance and infection control and its importance to RDVMs and clients is promote d in various ways. At the UF LAH, educational efforts about the program include provision of an infection control brochure to all new clients to educate them about the infection control measures that are instituted at the hospital to optimize patient care. In addition, every time a patient tests positive for Salmonella, the attending clinician immediately updates the affected cli ents and RDVMs about the patient Salmonella shedding status and infection control measures that were implemented in the hospital for that patient. At the time of discharge, clients with Salmonella positive horses are given a Salmonella fact sheet by the at tending clinician. The purpose of the fact sheet is to educate the client about the potential risks of having a Salmonella positive horse on a farm and measures that

PAGE 159

159 should be implemented at the farm level to reduce the risk of exposure to humans and other farm animals. One element of an animal disease surveillance system is evaluation and feedback. Evaluation of the performance of a hospital surveillance program may involve assessing the frequency of nosocomial infections within a given period of time, fr equency of environmental contamination with selected pathogens, compliance level by hospital clinicians and personnel, and level of awareness about the program among RDVMs and clients who refer or bring patients to the hospital. To have an efficient progra m it is important that feedback is collected from all individuals involved in the program including clinicians, hospital personnel, RDVMs and clients. Considering the role played by RDVMs and clients, it is important to know the level of awareness and per ceptions about the relevance of a hospital surveillance and infection program among this group. This information can be used to guide decision making on hospital policy issues related to surveillance and infection control and for streamlining costs and eff iciency of services provided to clients. The objective of this study was to assess the awareness and relevance of a hospital surveillance and infection control program among referral veterinarians and clients who refer or send horses to a referral hospita l for veterinary care. Materials and Methods Study Population All clients (n= 3095) who sent at least one equine patient to the UF LAH for veterinary care during July 2007 through July 2011 and whose contact information (mail/email address) was available were considered for inclusion in the study. Contact information (mail/email address) for eligible clients was obtained from hospital records

PAGE 160

160 for clients that brought equine patients to the UF LAH during July 2007 through July 2011. All RDVMs (n= 567) who referred equine patients to UF LAH in the past and whose contact information (mail/email address) was available were considered for inclusion in the study. Contact information (mail/email address) for eligible RDVMs was obtained from a list of RDVM contac ts kept by the UF LAH and only RDVMs that referred horses to the LAH in the past were eligible. Clients and RDVMs with incorrect or ineligible contact information were excluded. This study protocol was approved by the Institutional Review Board at the Univ ersity of Florida (Protocol # 2011 U 0729). Study Design A n electronic and mail survey was conducted to assess the awareness and relevance of the UF LAH surveillance and infection control program among referral veterinarians and clients who refer or send h orses to the UF LAH for veterinary care. Because surveys generally have low response rates, a convenience sampling approach was used to select eligible R DVMs and clients, and all selected study participants were sent surveys. Two separate 2 page questionn aires were designed for RDVMs and clients (Appendix A & B ). The RDVM and clients surveys consisted of 15 and 14 questions respectively, most of which were closed ended, and both surveys required at most ten minutes to complete. The RDVM questionnaire was p retested by administering questionnaires to 100 RDVMs during the UF Veterinary Hospitals referring veterinarian appreciation day on June 25 th 2011. Thirty five of 100 questionnaires were completed. Based on the pretest responses, questions in the survey in struments were revised to improve clarity. The client questionnaire was not pretested.

PAGE 161

161 The electronic (email) survey was sent on September 9, 2011 to all enrolled RDVMs (n=242) and clients (n=483) that had email addresses. The electronic correspondence inc luded an introductory email and a web link to the survey. The introductory letter explained the purpose of the study and assured the respondents that identifying information and responses would remain confidential. A remainder email communication was sent 10 days later to urge the respondents to complete the survey. After the second email communication, data from all respondents were entered into an excel database for analysis. Personal identifying information (names) was requested in the electronic surveys and this was delinked by assigning codes to each respondent prior to data entry. The mail survey was mailed out on November 21, 2011 to all enrolled R DVMs (n=225) and clients (2612) that did not participate in the electronic survey, or that did not have an email address but had a physical mailing address (and were never contacted during the time when the electronic survey was administered). A cover letter and a postage paid return envelope were included in the mailing. The cover letter described the purpo se of the study and assured the respondents that identifying information and responses would remain confidential. Respondents remained anonymous. To avoid duplication of respondents, identifying information (codes) was printed on each mail survey and used to identify the respondents during data analysis. One month was allowed for responses to be returned by mail. Remainder mail communication was not sent. Data for the respondents were entered into an excel database for analysis. Data Collection Briefly, the survey instrument for RDVMs focused on three major areas and respondents were asked questions on the following variables (Appendix A ): (1) general

PAGE 162

162 information: county in which practice is based; years in practice; type of clinical practice; animal species seen at practice; number of horses referred to the UF Hospital for veterinary care in the last two years; (2) awareness: awareness that the UF LAH operates a surveillance and infection control program; awareness that the UF LAH has an infection control c ommittee; awareness that the UF LAH has an infection control officer; awareness that horses presenting with signs of gastrointestinal tract disease are sampled and tested for diagnosis of Salmonella shedding during hospitalization; awareness that horses wi th diarrhea, fever and leucopenia, or that test positive to Salmonell a shedding at admission or during hospitalization are placed in isolation; awareness that every time there is evidence that a horse has potentially acquired a nosocomial Salmonella infect ion during hospitalization, enhanced infection control measures are implemented immediately; (3) relevance: importance of being informed by a UF Hospital attending clinician when a horse (that you referred) tests positive to Salmonella shedding; importanc e of sampling and testing horses for Salmonella shedding; and importance of a referral hospital operating a surveillance and infection control program. The survey instrument for clients focused on three major areas and respondents were asked questions on t he following variables (Appendix B ): (1) general information: county in which horse farm/premises is located; number of horses on premises; main use of horses on premises; number of horses brought to the UF Hospital for veterinary care in the last two year s; (2) awareness: awareness that the UF LAH operates a surveillance and infection control program; awareness that the UF LAH has an infection control committee; awareness that the UF LAH has an infection control officer;

PAGE 163

163 awareness that horses presenting w ith signs of gastrointestinal tract disease are sampled and tested for diagnosis of Salmonella shedding during hospitalization; awareness that horses with diarrhea, fever and leucopenia, or that test positive to Salmonell a shedding at admission or during h ospitalization are placed in isolation; awareness that every time there is evidence that a horse has potentially acquired a nosocomial Salmonella infection during hospitalization, enhanced infection control measures are implemented immediately; (3) relevan ce: importance of being informed by a UF Hospital attending clinician when a horse (that you referred) tests positive to Salmonella shedding; importance of sampling and testing horses for Salmonella shedding; and importance of a referral hospital operatin g a surveillance and infection control program. Data Analysis Descriptive statistics were determined for each variable o f interest. Frequency distributions were determined for the categorical variables (yes/no) Medians, first quartiles, and third quartile s were calculated for continuos variables (e.g., number of horses on farm, number of practice years). C omparison s between continuos variables were made using the wilcoxon rank sum test (e.g., number of years in practice for RDVMs and whether they found tes ting to be expensive (yes/no) ) Comparisons between categorical variables were made using chisquare. Results Overall, 92 of 567 ( 16 %) surveys were returned by RDVMs and 594 of 3095 ( 19 %) were returned by clients during November 2011 through January 2012. E lectronic survey responses from 18/242 RDVMs and 95/483 clients were received. M ail survey responses from 39/ 225 RDVMs and 499 /2612 clients were received. Following

PAGE 164

164 pretesting of the R DVM questionnaire in the early phase of the study 35/100 responses were received and these data were added to data from the electronic and mail surveys in the analysis Response trend for the mail surveys was as follows. Of the 39 mail responses from RDVMs, 2 responses were received one week after mailing, 22 responses afte r two weeks, 7 responses after 3 weeks, 5 responses after 4 weeks, and the remaining 3 responses were received between weeks 5 7. Of the 499 mail responses from clients, 51 responses were received one week after mailing, 269 responses after two weeks, 94 r esponses after 3 weeks, 33 responses after 4 weeks, and the remaining 52 responses were received between weeks 5 10. General Information for Referral D VM s Most RDVMs who responded to the survey had their practiced based in Marion county (17/92 or 22%) (Tab le 5 1). The median number of years in practice was 22 (first quartile = 10; third quartile = 30). The median number of horses referred by RDVMs to the UF LAH in the last two years was 5 (2, 10). Most RDVMs classified their practiced General Information for Clients Among clients who responded to this survey, the main location of horse farms in Florida was Marion C ounty (69/594 or 12 %) (Table 5 2). The main single use of horses 94 or 13%). The median number of horses on the premises was 5 (2, 10), and the median number of horses brought to the UF LAH in the last two years for veterinary care was 1 (1, 1).

PAGE 165

165 Awareness Among RDVMs and clients who responded to the survey, more RDVMs (68%) were aware that the UF LAH operates a surveillance and infection control program, compared to clients (39%) ( P < 0.05) (Table 5 3). Relevance Almost all RDVMs (97%) and clients (97%) considered it important to be informed by the UF Hospital attendin g clinician when a horse (their patient) tested positive to Salmonella shedding at admission or during hospitalization (Table 5 4). Both RDVMs (99%) and clients (96%) considered testing of horses with colic or diarrhea for early detection of Salmonella she dding in feces upon admissi on and during hospitalization t o be justified. However, more clients (25%) considered the cost of testing (~ $56) to be expensive, compared to RDVMs (11%) ( P < 0.05). The number of horses owned by clients was similar among client s who considered the cost expensive (Median = 5 horses; 3, 12), compared to those that did not (median = 4 horses; 2, 10). The median number of years in practice was lower among RDVMs that found cost of testing expensive (median = 9 years; 5 1 5) compared to RDVMs that did not ( 24 ; 12, 30) ( P = 0.01). Using a sca le of 1 (not important) to 10 (very important), both RDVMs (median = 10; 1 st quartile = 8; 3 rd quartile = 10) and clients (median = 10; 8, 10) considered it very important that a referral hospital operates a surveillance and infection control program. Finally, most RDVMs (88%) and clients (82%) responded that if the UF LAH were to experience an outbreak of salmonellosis and was forced to close temporarily for cleaning and disinfection of hospital f acilities, they would consider referring horses to the hospit al again after re opening

PAGE 166

166 Discussion In this study, a survey was conducted to assess awareness and relevance of a hospital surveillance and infection control program among referral veterinarian s and clients who refer or send horses to a referral hospital for veterinary care. Overall, most (7 of every 10) RDVMs were aware that the referral hospital operates a surveillance and infection control p rogram. In contrast, few clients (4 of every 10) wer e aware B oth R DVMs and clients indicated that it is very important that a referral hospital operates a surveillance and infection control program. Among RDVMs and clients who responded to the survey 7 of every 10 RDVMs were aware that the referral hospit al operates a surveillance and infection control program, compared to only 4 of every 10 clients. Currently, educational efforts targeting clients at the UF LAH include provision of an infection brochure to new clients and a Salmonella fact sheet to client s with Salmonella positive horses Th e finding that only 4 of every 10 clients are aware of the program suggests that more education efforts are required. Almost all RDVMs and clients considered it important to be informed by the h ospital attending clinic ian when a horse tested positive to Salmonella shedding at admission or during hospitalization In addition, both RD V M s and clients c onsidered testing of horses with colic or diarrhea for early detection of Salmonella shedding in feces upon admission and d uring hospitalization to be justified. Th ese finding s indicate that RDVMs and clients understand the potential benefits to patient care of having a hospital surveillance program. In addition, these findings suggest that clinician RDVM/client interaction ma y be an important avenue for creating awareness about the program.

PAGE 167

167 M ore clients (2 to 3 of every 10) considered the cost of testing to be expensive, compared to RDVMs (1 of every 10) At the UF LAH, c lients pay for hospital charges related to testing for s urveillance purposes and for isolation stall fees. Cost of testing and isolation varies depending on duration of hospitalization. On average, horses that present with signs of gastrointestinal disease are hospitalized for 3 4 days and two fecal samples are collected and submitted for bacteriological culture testing, resulting in a cost of approximately $56 ($28 per sample). I solation stall fees for patients that test Salmonella positive or have diarrhea or fever and leucopenia are $115 per day In this stud y, it is not known if the demographics of clients who considered cost of testing to be expensive were different from those who did not. A mong RDVMs, number of y ears in practice ( 9 years) was associated with the respon se that cost of testing was expensive. B oth RDVMs and clients considered it very important that a referral hospital operates a surveillance and infection control program and most RDVMs and clients responded that i f the hospital were to experience an outbreak of salmonellosis and was forced to close temporarily for cleaning and disinfection of hospital facilities, they would consider referring horses to the hospital again after re opening. This finding again suggest s th at RDVMs and clients appreciate the benefits of having a hospital surveillance and infection control program and efforts that are aimed at optimizing patient care. The results of this study were subject to a number of limitations. First, the response rates were low (16% for RDVMs and 19% for clients) and this may have resulted in non response bias. In this study, follow up contact to solicit for responses

PAGE 168

168 was only performed for electronic surveys but not mail surveys due to financial constraints. It i s not known if the demographics of RDVMs and clients that responded were different from those who chose not to participate. It is also not known if a higher response rate in the present study would have provided different results. In conclusion, this stud y revealed that (1) a high frequency of clients who responded to this survey were not aware that the referral hospital operates a surveillance and infection control program to reduce the risk of hospital acquired infections caused by pathogens such as Salm onella ; (2) m ost RDVMs and clients find the frequency of testing for Salmonella justified and not expensive; and (3) both RDVMs and clients considered it very important that a referral hospital operates a surveillance and infection control program.

PAGE 169

169 Table 5 1. Survey responses from RDVMs: general information Variable RDVMs N = 92 (100%) In what county is your practice based? Marion Alachua Lake Hillsborough Other 17 (22) 7 (9) 7 (9) 4 (5) 43 (55) What type of clinical practice do you wo rk in? Mobile Services only Clinic & Mobile Clinic Services only Teaching/ Referral Hospital Other 47 (51) 35 (38) 7 (8) 2 (2) 1 (1) What animal species do you see at your practice? Equine Exclusive Large Animal, all species Mi xed Practice Other 51 (55) 20 (22) 17 (19) 4 (4) How many years have you practiced veterinary medicine? 22 (10, 30)* In the last two years (eg, summer 2009 to summer 2011), approximately how many horses have you referred to the UF Hospital for veterinary care? 5 (2, 10)* Other = counties where the number of RDVMs < 4. ** Data presented as median (1 st quartile, 3 rd quartile).

PAGE 170

170 Table 5 2. Survey responses from clients: general information Variable Clients N = 594 (100%) In what county is y our horse farm/premises located? Marion Alachua Volusia Lake Duval Clay Brevard St. Johns Other 69 (12) 41 (7) 31 (5) 26 (4) 26 (4) 23 (4) 23 (4) 23 (4) 325 (55) What is the main use of horses that you have on your premises? Pleasure Jumper/ Hunter Show/ Dressage Riding (Trail/ Hiking) Pleasure and Riding (Trail/Hiking) Pet Breeding Racing Breeding and Racing Other ** 73 (13) 44 (8) 39 (7) 28 (5) 24 (4) 20 (3) 18 (3) 14 (2) 14 (2) 311 (53) How many horses do you have on your premises? 5 (2,10)* ** In the last two years (eg, summer 2009 to summer 2011), approximately how many horses have you brought to the UF Hospital for veterinary care? 1 (1, 1)* ** Other = the number of clients in a co unty < 23 ** Other = the number of horses < 14 *** Data presented as median (1 st quartile, 3 rd quartile).

PAGE 171

171 Table 5 3 Survey responses from RDVMs and clients: awareness Variable RDVMs N = 92 (100%) Clients N = 594 (100%) Did you know the UF LAH operates a surveillance and infection control program to reduce the risk of hospital acquired infections caused by pathogens such as Salmonella? No Yes 29 (32) 62 (68) 358 (61) 227 (39) Variable RDVMs N = 62 (100%) Clients N = 227 (100%) Were y ou aware that the UF Large Animal Hospital has an infection control committee that meets quarterly (or more often) to assess the overall hospital infection control status? No Yes 45 (75) 15 (25) 162 (76) 52 (24) Were you aware that the UF LAH has an infection control officer that coordinates day to day surveillance and infection control activities? No Yes 39 (65) 21 (35) 151 (71) 63 (29) Did you know that horses presenting with signs of gastrointestinal tract disease are sampl ed and tested for diagnosis of Salmonella shedding at admission and during hospitalization? No Yes 4 (7) 57 (93) 87 (41) 128 (59) Did you know that horses with diarrhea, fever and low white blood cell count, or that test positive to Salmon ella shedding at admission or during hospitalization are placed in isolation? No Yes 0 (0) 61 (100) 61 (28) 154 (72) Did you know that every time there is evidence that a horse has potentially acquired a hospital acquired Salmonella infe ction during hospitalization, enhanced infection control measures are implemented immediately? No Yes 7 (12) 54 (88) 57 (27) 158 (73)

PAGE 172

172 Table 5 4. Survey responses from RDVMs and clients: relevance Variables RDVMs N = 92 (100%) Clients N = 594 (100%) Is it important for you to be informed by the UF Hospital attending clinician when a horse (your patient) tests positive to Salmonella shedding at admission or during hospitalization? No Yes 3 (3) 86 (97) 16 (3) 570 (97) O n average, horses presenting with signs of gastrointestinal tract disease at the UF hospital are hospitalized for 5 days. This group of horses is sampled and tested for diagnosis of two fecal samples. The cost of testing to the client is $28 per sample o r $56 for both samples? Do you find this level of testing justified? No Yes Do you find this cost expensive? No Yes 1 (1) 90 (99) 81 (89) 10 (11) 20 (4) 554 (96) 428 (75) 144 (25) Some clients have expressed that they would no t consider sending their horses to a referral hospital that does not operate a surveillance and infection control program because of the perceived risk of disease transmission. Do you feel the same way? On a scale of 1 (not important) to 10 (very importa nt), how important is it for you that a referral hospital operates a surveillance and infection control program? 10 (8, 10)* 10 (8, 10)* In the past, several veterinary hospitals have been forced to close temporarily for 1 to 3 months for cleaning and disinfection because of Salmonella outbreaks in an even t to occur in our hospital, if the UF LAH were to experience an outbreak of salmonellosis and was forced to close temporarily for c leaning and disinfection of hospital facilities, would you consider referring horses to the hospital again after re opening? No Yes I am not sure 5 (6) 80 (88) 6 (6) 18 (3) 480 (82) 85 (15) *Data presented as median (1 st quart ile, 3 rd quartile).

PAGE 173

173 CHAPTER 6 SUMMARY AND RESEARCH CONCLUSIONS This research work addressed four objectives. First, this research work provide d a critical review and analysis of published studies that have contributed to the literature of epidemiology and infection control aspects of nosocomial Salmonella infections in hospitalized horses. Th e review was structured in three main parts: surveillance and infection control; epidemiological research; and a discussion of relevant infection control issues and ar eas of research that can further improve existing hospital surveillance and infection control programs. This critical review identif ied knowledge gaps in the epidemiology of Salmonella infections in hospitalized horses, some which were addressed in this re search work. Second, an investigation of the relationships between clinical signs, hematological and plasma chemistry parameters, and clinical procedures, and Salmonella shedding in GI horses with or without diarrhea was conducted. Study results indicated that a mong horses that tested positive for Salmonella use of antimicrobials and anti inflammatories before admission, high plasma triglycerides, low plasma sodium, low plasma protein, and high mean corpuscular hemoglobin concentration at admission were associated wi th Salmonella shedding. Among horses that tested positive for Salmonella diarrhea, high plasma protein concentration and high plasma triglycerides at admission were associated with Salmonella shedding. Among horses with out diarrhea that tested positive for Salmonella associated with Salmonella shedding. In conclusion, high plasma protein concentration high plasma triglycerides abdominal surgery and season are parameters that can be

PAGE 174

174 considered for evaluation in hospital biosecurity standard operating procedures to identify equine GI inpatients at high risk of Salmonella shedding. The third research objective was to assess the diagnostic performance of a real ti me PCR assay for detection of Salmonella spp in feces of hospitalized horses, compared to bacteriological culture. Results i ndicated that the o ver all relative specificity of the PCR assay when using primer set 1 in horses without signs of GI disease that t ested negative and horses with signs of GI disease that tested negative was 87 to100% and 88 to 98 % respectively. The o ver all relative specificity of the PCR assay when using primer set 2 in horses without signs of GI disease that tested negative and horse s with signs of GI disease that tested negative was 97 to100% and 86 to 95 % respectively The overall relative sensitivity of the assay when using primer set 1 or 2 in horses with signs of GI disease that tested positive was 100%. Based on the relative spe cificity and sensitivity estimates determined using horses classified as true negatives and true positives this real time PCR protocol ( using primer set 1 or 2 ) is reliable, with a relatively high specificity and sensitivity compared to bacteriological cu lture. This real time PCR protocol can be used as a surveillance tool to detect Salmonella spp in fecal specimen of hospitalized horses. Considering its advantage of being rapid (reporting laboratory results approximately 48 hours later), compared to cultu re, this PCR assay can be used as a screening test that may be confirmed by culture if a samples tests PCR positive. In addition, culture would provide additional data (a ntibiogram and serotyping data ) that are needed to establish if there is an epidemiolo gical relationship between Salmonella cases or isolates, thus providing evidence that Salmonella infection or colonization is nosocomial in origin.

PAGE 175

175 The fourth objective was to assess the awareness and relevance of a hospital surveillance and infection con trol program among referral veterinarians and clients who refer or send horses to a referral hospital for veterinary care Survey results revealed that (1) most clients who responded to this survey were not aware that the referral hospital operates a surve illance and infection control program to reduce the risk of hospital acquired infections caused by pathogens such as Salmonella ; (2) Most RDVMs and clients find the frequency of testing for Salmonella justified and not expensive; and (3) both RDVMs and cli ents considered it very important that a referral hospital operates a surveillance and infection control program. This research work addressed four important aspects that are relevant to the epidemiology, diagnostics, and surveillance and infection control of Salmonella infections in hospitalized horses. First, a critical review of the literature of the epidemiology and infection control aspects of nosocomial Salmonella infections in hospitalized horses was performed. This critical review identified several knowledge gaps, three of which were addressed in this research wo rk. Second, a n investigation of the relationships between clinical signs, hematological and plasma chemistry parameters, and clinical procedures, and Salmonella shedding in GI horses with or without diarrhea was performed F indings from this investigation can be considered for use in current hospital surveillance programs to improve early detection of Salmonella shedding and permit rapid implementation of infection control measures. Third, a n investigation was performed to assess the diagnostic performance of a real time PCR assay for detection of Salmonella spp in feces of hospitalized horses, compared to bacteriological culture. Findings from this investigation revealed that the PCR protocol

PAGE 176

176 that was evaluated was reliable, with a relatively high specificity and sensitivity compared to bacteriological culture. This real time PCR protocol can be used as a surveillance tool to detect Salmonella spp in fecal specimen of hospitalized horses. Fina lly a survey was conducted to assess the awareness and relevance of a hospital surveillance and infection control program among referral veterinarians and clients. Findings from this s urvey can be used to justify the need for enhance d education efforts abo ut the program among clients guide decision making on hospital policy issues related to surveillance and infection control and streamlin e costs and efficiency of services provided to clients.

PAGE 177

177 APPENDIX A SURVEY FOR R EFERRAL DVM S S U R V E Y edback is appreciated! Please answer the questions in sections A, B, and C. This survey includes 15 questions and will take no more than 10 minutes of your time to complete. Thank you for your participation! SECTION A: GENERAL INFORMATION 1. In what county is your practice based? 2. How many years have you practiced veterinary medicine? [ ] 3. What type of clinical practice do you work in? Teaching/referral hospital [ ] Mobile services only [ ] Clinic services only [ ] Clinic & mobile [ ] Other (pleas e specify) 4. What animal species do you see at your practice? Equine exclusive [ ] Bovine exclusive [ ] Large animal, all species [ ] Feline exc lusive [ ] Small animal, a ll species [ ] Mixed practice [ ] Other (please specify) 5. In the last two y ears (eg, summer 2009 to summer 2011), approximately how many horses have you referred to the UF Hospital for veterinary care? [ ] 6. Did you know that the UF Large Animal Hospital operates a surveillance and infection control program to reduce the risk of hospital acquired infections caused by pathogens such as Salmonella ? No [ ] Yes [ ] If your answer to this question is No, please continue with question No. 12 in section C. SECTION B: AWARENESS 7. Were you aware that the UF Large Animal Hospital has an i nfection control committee that meets quarterly (or more often) to assess the overall hospital infection control status? No [ ] Yes [ ] 8. Were you aware that the UF Large Animal Hospital has an infection control officer that coordinates day to day surveill ance and infection control activities under the supervision of a hospital epidemiologist? No [ ] Yes [ ] 9. Did you know that horses presenting with signs of gastrointestinal tract disease are sampled and tested for diagnosis of Salmonella shedding at admis sion and during hospitalization? No [ ] Yes [ ]

PAGE 178

178 10. Did you know that horses with diarrhea, fever and leukopenia, or that test positive to Salmonell a shedding at admission or during hospitalization are placed in isolation? No [ ] Yes [ ] 11. Did you know that every time there is evidence that a horse has potentially acquired a nosocomial Salmonella infection during hospitalization, enhanced infection control measures are implemented immediately (eg, footmats with disinfectant, use of gloves and gowns are mandat ory on every large animal inpatient)? No [ ] Yes [ ] SECTION C: RELEVANCE 12. Is it important for you to be informed by the UF Hospital attending clinician when a horse (that you referred) tests positive to Salmonella shedding at admission or during hospita lization? No [ ] Yes [ ] 13. On average, horses presenting with signs of gastrointestinal tract disease at the UF Hospital are hospitalized for 5 days. This group of horses is sampled and tested for diagnosis of Salmonella shedding at admission and an addit ional time during hospitalization for a total of two fecal samples. The cost of testing to the client is $28 per sample or $56 for both samples. Do you find this level of testing justified? No [ ] Yes [ ] Do you find this cost expensive? No [ ] Yes [ ] 14. Some clients have expressed that they would not consider sending their horses to a referral hospital that does not operate a surveillance and infection control program because of the perceived risk of disease transmission. Do you feel the same way? On a scale of 1 (not important) to 10 (very important), how important is it for you that a referral hospital operates a surveillance and infection control program? 1[ ] 2[ ] 3[ ] 4[ ] 5[ ] 6[ ] 7[ ] 8[ ] 9[ ] 10[ ] 15. In t he past, several veterinary hospitals in the US have been forced to close temporarily for 1 to 3 months for cleaning and disinfection because of Salmonella outbreaks in l, if the UF Large Animal Hospital were to experience an outbreak of salmonellosis and was forced to close temporarily for cleaning and disinfection of hospital facilities, would you consider referring horses to the hospital again after re opening? No [ ] Yes [ ] I am not sure [ ] If you have additional comments, please use the space below to share your feedback. Thank you again for your participation!

PAGE 179

179 APPENDIX B SURVEY FOR CLIENTS S U R V E Y Please answer the que stions in sections A, B, and C. This survey includes 14 questions and will take no more than 10 minutes of your time to complete. Thank you for your participation! SECTION A: GENERAL INFORMATION 1. In what county is your horse farm/premises located? 2. How many horses do you have on your premises? [ ] 3. What is the main use of horses that you have on your premises? Pleasure [ ] Pet [ ] Racing [ ] Show/dressage ` [ ] Jumper/Hunter [ ] Riding (trail, hiking) [ ] Breeding [ ] Retired [ ] Ropin g [ ] Other (please specify) 4. In the last two years (eg, summer 2009 to summer 2011), approximately how many horses have you brought to the UF Hospital for veterinary care? [ ] 5. Did you know that the UF Large Animal Hospital operates a surveillance a nd infection control program to reduce the risk of hospital acquired infections caused by pathogens such as Salmonella ? No [ ] Yes [ ] If your answer to this question is No, please continue with question No. 11 in section C. SECTION B: AWARENESS 6. Were y ou aware that the UF Large Animal Hospital has an infection control committee that meets quarterly (or more often) to assess the overall hospital infection control status? No [ ] Yes [ ] 7. Were you aware that the UF Large Animal Hospital has an infection c ontrol officer that coordinates day to day surveillance and infection control activities under the supervision of a hospital epidemiologist? No [ ] Yes [ ] 8. Did you know that horses presenting with signs of gastrointestinal tract disease are sampled and t ested for diagnosis of Salmonella shedding at admission and during hospitalization? No [ ] Yes [ ] 9. Did you know that horses with diarrhea, fever and low white blood cell count, or that test positive to Salmonell a shedding at admission or during hospitali zation are placed in isolation?

PAGE 180

180 No [ ] Yes [ ] 10. Did you know that every time there is evidence that a horse has potentially acquired a hospital acquired Salmonella infection during hospitalization, enhanced infection control measures are implemented imme diately (eg, footmats with disinfectant, use of gloves and gowns are mandatory on every large animal inpatient)? No [ ] Yes [ ] SECTION C: RELEVANCE 11. Is it important for you to be informed by the UF Hospital attending clinician when a horse (your patient ) tests positive to Salmonella shedding at admission or during hospitalization? No [ ] Yes [ ] 12. On average, horses presenting with signs of gastrointestinal tract disease at the UF Hospital are hospitalized for 5 days. This group of horses is sampled and tested for diagnosis of Salmonella shedding at admission and an additional time during hospitalization for a total of two fecal samples. The cost of testing to the client is $28 per sample or $56 for both samples. Do you find this level of testing justif ied? No [ ] Yes [ ] Do you find this cost expensive? No [ ] Yes [ ] 13. Some clients have expressed that they would not consider sending their horses to a referral hospital that does not operate a surveillance and infection control program because of the p erceived risk of disease transmission. Do you feel the same way? On a scale of 1 (not important) to 10 (very important), how important is it for you that a referral hospital operates a surveillance and infection control program? 1[ ] 2[ ] 3[ ] 4[ ] 5[ ] 6[ ] 7[ ] 8[ ] 9[ ] 10[ ] 14. In the past, several veterinary hospitals in the US have been forced to close temporarily for 1 to 3 months for cleaning and disinfection because of Salmonella outbreaks in horses or food an if the UF Large Animal Hospital were to experience an outbreak of salmonellosis and was forced to close temporarily for cleaning and disinfection of hospital facilities, would you con sider referring horses to the hospital again after re opening? No [ ] Yes [ ] I am not sure [ ] If you have additional comments, please use the space below to share your feedback. Thank you again for your participation!

PAGE 181

181 LIST OF REFERENCES 1. H ird DW, Pappaioanou M Smith BP. Case control study of risk factorsvassociated with isolation of Salmonella S aintpaul in hospitalized horses. Am J Epidemiol 1984; 120:852 864. 2. Tillotson K, Savage CJ, Salman MD et al. Outbreak of Salmonella infantis inf ection in a large animal veterinary teaching hospital. J Am Vet Med Assoc 1997; 211: 1554 1557. 3. Schott HC, Ewart SL, Walker RD, et al An outbreak of salmonellosis among horses at a veterinary teaching hospital. J Am Vet Med Assoc 2001;218:1151 1159. 4 Ward MP, Brady TH, Couetil LL, et al Investigation and control an outbreak of Salmonella typhimurium in a population of hospitalized horses. Vet Microbiol 2005;107:233 240. 5. Dallap Schaer BL, Aceto H, Rankin SC. Outbreak of salmonellosis caused b y Salmonella enterica serovar Newport MDR AmpC in a large animal veterinary teaching hospital. J Vet Intern Med 2010;24:1138 46. 6. Ekiri AB, Morton AJ, Long MT, et al Review of the epidemiology and infection control aspects of nosocomial Salmonella in fections in hospitalized horses. Equine Vet Educ 2010;22:631 641. 7 Cohen ND, Neibergs HL, Wallis DE, et al. Genus specific detection of salmonellae in equine feces by use of the polymerase chain reaction. Am J Vet Res 1994;8:1049 1054. 8 Ward MP, A linovi CA, Couetil LL et al Evaluation of a PCR to detect Salmonella in fecal samples of horses admitted to a veterinary teaching hospital. J Vet Diagn Invest 2005;17:118 123. 9 Gentry Weeks C, Hutcheson HJ, Kim LM, et al. Identification of two phyl ogenetically related organisms from feces by PCR for detection of Salmonella spp. J Clin Microbiol 2002;4 :1 487 1492 10. Pusterla N, Byrne BA, Hodzic E, et al. Use of quantitative real time PCR for the detection of Salmonella spp. in fecal samples from ho rses at a veterinary teaching hospital. Vet J 2010;186(2):252 5. 1 1 Cohen ND, Martin LJ, Simpson RB, et al. Comparison of polymerase chain reaction and microbiological culture for detection of salmonellae in equine feces and environmental samples. Am J Vet Res 1996; 6:780 786.

PAGE 182

182 1 2 Kurowski PB, Traub Dargatz JL, Morley PS et al Detection of Salmonella spp in fecal specimens by use of real time polymerase chain reaction assay. Am J Vet Res 2002;6 3 :1265 1268. 13. Bohaychuk VM, Gensler GE, McFall ME, et al. A real time PCR assay for the detection of Salmonella in a wide variety of food and food animal matrices. J Food Prot 2007;5:1080 1087. 14 Farmer JJ. Enterobacteriaceae: Introduction and identification. In: Murray PR, Baron EJ, Pfaller MA, eds. Manual of clinical m icrobiology 6th ed. Washington, DC: American Society for Microbiology, 1995;438 449. 1 5 Kahn CM and Line S. Salmonellosis. In: Gyles CL, ed. The Merck Veterinary Manual Whitehouse station, New Jersey: MERCK & Co, 2005;157 234. 1 6 Smith BP. Salmonello sis. In: Smith BP, ed. Large animal internal medicine St Louis:CV Mosby Co, 1991;818 822. 1 7 Spiers SJ. Salmonellosis. Vet Clin North Am Equine Pract 1993;9:385 397. 18. Madic J, Hajsig B, Sostaric B, et al. An outbreak of abortion in mares associate d with Salmonella abortusequi infection. Equine Vet J 1997;29:203 233. 1 9 Mastroeni P, Chabalgoity JA, Dunstan SJ, et al. Salmonella : Immune Responses and Vaccines. Vet J 2000;161:132 164. 20 Michetti P, Mahan MJ, Slauch JM, et al. Monoclonal secretory immunoglobulin A protects against oral challenge with the invasive pathogen Salmonella typhimurium Infect Immun 1992;60:1786 1792. 21 Santos RL, Zhang S, Tsolis RM, et al. Morphologic and molecular characterization of Salmonella typhimurium infection i n neonatal calves. Vet Pathol 2002;39:200 215. 22. Jones BD, Ghori N Falkow S. Salmonella typhimurium initiates murine infection patches. J Exp Med 1994;180:15 23. 23 Sant os RL, Zhang S, Tsolis RM, et al. Morphologic and molecular characterization of Salmonella typhimurium infection in neonatal calves. Veterinary Pathology 2002 ; 39: 200 215. 24. Jones SL. Inflammatory diseases of the gastrointestinal tract causing diarrhea. In: Reed SM, Bayly WM, Sellon DC, eds. Equine Internal Medicine. 2nd ed. St Louis, Missouri, USA: Saunders 2004;884 913.

PAGE 183

183 25 Giannella RA. Pathogenesis of acute bacterial diarrheal disorders. Annu Rev Med 1981;32:341 357. 26 Ohl ME, Miller SI. Salmonella : a model for bacterial pathogenesis. Annu Rev Med 2001;52:259 274. 2 7 Santos RL, Tsolis RM, Baumler AJ, et al. Pathogenesis of Salmonella induced enteritis. B J Med Biol Res 2003;36:1 12. 2 8 Karem KL, Kanangat S Rouse BT. Cytokine expression in the g ut associated lymphoid tissue after oral administration of attenuated Salmonella vaccine strains. Vaccine 1996;14:1495 1502. 2 9 Koo FC, Peterson JW, Houston CW et al. Pathogenesis of experimental salmonellosis: inhibition of protein synthesis by cytotox in. Infect Immun 1984;43:93 100. 30 Giannella RA, Gots RE, Charney AN et al. Pathogenesis of Salmonella mediated intestinal fluid secretion: activation of adenylate cyclase and inhibition by indomethacin. Gastroenterology 1975;69:1238 1245. 31 Peterso n JW, Molina NC, Houston CW et al. Elevated cAMP in intestinal epithelial cells during experimental cholera and salmonellosis. Toxicon 1983;21:761 775. 32 Chopra AK, Huamg JH, Xu X et al. Role of Salmonella enterotoxin in overall virulence of the organ ism. Micro Pathog 1999;27:155 171. 33. Watson PR, Galyov EE, Paulin SM et al. Mutation of invH, but not stn, reduces Salmonella induced enteritis in cattle. Infect Immun 1998;66:1432 1438. 34 Giannella RA. Importance of the intestinal inflammatory reaction in Salmonella mediated intestinal secretion. Infect Immun 1979;23:140 145. 35 Galan JE, Collmer A. Type III secretion machines: bacterial devices for protein delivery into host cells. Scien ce 1999;284:1322 1328. 36 changes in intestinal structure and function. J Pediatr Gastroenterol Nutr 1991;12:5 20. 3 7 Murray MJ. Digestive physiology of the large intestine in ad ult horses. Part 2. Pathophysiology of colitis. Compend Cont Educ Pract Vet 1988;10:1309 1316. 3 8 Powell DW. Neuroimmunophysiology of the gastrointestinal mucosa: implications for inflammatory diseases. Tran Am Clin Climatol Assoc 1994;106:124 138.

PAGE 184

184 3 9 McCormick BA, Miller SI, Carnes D et al. Transepithelial signaling to neutrophils by salmonellae: a novel virulence mechanism for gastroenteritis. Infect Immun 1995;63:2302 2309. 40 Madara JL. Review article: pathobiology of neutrophil interactions with intestinal epithelia. Aliment Pharmacol Ther 2000;3 (suppl 11):57 62. 41 Groisman EA Ochman H. How Salmonella became a pathogen. Trends Microbiol 2007;5:343 349. 42 Hansen Wester I Hensel M. Salmonella pathogenicity islands encoding type III secreti on systems. Microbes Infect 2001;7:549 59. 43 Hueck CJ. Type III protein secretion systems in bacterial pathogens of animals and plants. Microbiol Mol Biol Rev 1998;62:379 433. 44 Hensel M. Salmonella Pathogenicity Island 2. Mol Microbiol 2000;36:1015 1024. 45 Wallis TS Galyov EE. Molecular basis of Salmonella induced enteritidis, Mol Microbiol 2000;36:997 1005. 4 6 Lee CA Falkow S. The ability of Salmonella to enter mammalian cells is affected by bacterial growth state. Proc Natl Acad Sci USA 199 0;87:4304 4308. 4 7 Bajaj V, Lucas RL, Hwang C, et al Co ordinate regulation of Salmonella typhimurium invasion genes by environmental and regulatory factors is mediated by control of hilA expression. Molecular Microbiolog y 1996;22:703 714. 4 8 Ahmer BM M, Van Reeuwijk J, Watson PR, et al. Salmonella SirA is a global regulator of genes mediating enteropathogenesis. Molecular Microbiology 1999;31: 971 982. 4 9 Watson PR, Paulin SM, Bland AP, et al Characterization of intestinal invasion by Salmonella typh imurium and Salmonella dublin and effect of a mutation in the invH gene. Infection and Immunity 1995;63:2743 2754. 50 Watson PR, Galyov EE, Paulin SM, et al. Mutation of invH but not stn reduces Salmonella induced enteritis in cattle. Infection and Imm unity 1998;66:1432 1438. 51. Tsolis RM, Adams LG, Ficht TA et al. Contribution of Salmonella typhimurium virulence factors to diarrheal disease in calves. Infection and Immunity 1999;67: 4879 4885. 52 Hayward RD Koronakis V. Direct nucleation and bund ling of actin by the SipC protein of invasive Salmonella EMBO Journal 1999;18:4926 4934.

PAGE 185

185 53 Zhou D, Mooseker MS Galn JE. Role of the S. typhimurium actin binding protein SipA in bacterial internalization. Science 1999;283:2092 2095. 54 McGhie EJ, Hay ward RD Koronakis V. Cooperation between actin binding proteins of invasive Salmonella : SipA potentiates SipC nucleation and bundling of actin. EMBO Journal 2001;20:2131 2139. 5 5 Stender S, Friebel A, Linder S, et al Identification of SopE2 from Salmon ella typhimurium a conserved guanine nucleotide exchange factor for Cdc42 of the host cell. Molecular Microbiology 2000 ; 36: 1206 1221. 5 6 Hardt WD, Chen LM, Schuebel KE, et al. S. typhimurium encodes an activator of Rho GTPases that induces membrane ruf fling and nuclear responses in host cells Cell 1998;93:815 826. 5 7 Fu Y and Galn JE. A Salmonella protein antagonizes Rac 1 and Cdc42 to mediate host cell recovery after bacterial invasion. Nature 1999;401: 293 297. 5 8 Kaniga K, Uralil J, Bliska JB et al A secreted protein tyrosine phosphatase with modular effector domains in the bacterial pathogen Salmonella typhimurium Mol Microbiol 1996;21: 633 641. 5 9 Eckmann L, Kagnoff MF Fierer J. Epithelial cells secrete the chemokine interleukin 8 in respon se to bacterial entry. Infection and Immunity 1993;61: 4569 4574. 60 Santos RL, Tsolis RM, Zhang S, et al. Salmonella induced cell death is not required for enteritis in calves. Infection and Immunity 2001;69:4610 4617. 61 McCormick BA, Hofman PM, Kim J, et al. Surface attachment of Salmonella typhimurium to intestinal epithelia imprints the subepithelial matrix with gradients chemotactic for neutrophils. Journal of Cell Biology 1995;131:1599 1608. 62 McCormick BA, Parkos CA, Colgan SP, et al. Apical secretion of a pathogen elicited epithelial chemoattractant activity in response to surface colonization of intestinal epithelia by Salmonella typhimurium Journal of Immunology 1998;160:455 466. 63 Hobbie S, Chen LM, Davis RJ et al. Involvement of mito gen activated protein kinase pathways in the nuclear responses and cytokine production induced by Salmonella typhimurium in cultured intestinal epithelial cells. Journal of Immunology 1997;159:5550 5559. 64 Santos RL, Zhang S, Tsolis RM, et al Animal mo dels of Salmonella infections: gastroenteritis vs typhoid fever. Microbes and Infection 2001;3:1335 1344.

PAGE 186

186 6 5 McCormick BA, Miller SI, Delp Archer C, et al Transepithelial signaling to neutrophils by Salmonella : a novel virulence mechanism for gastroenter itis. Infect Immun 1995; 63 : 2 302 2309. 6 6 Galyov EE, Wood MW, Rosqvist R, et al. A secreted effector protein of Salmonella dublin is translocated into eukaryotic cells and mediates inflammation and fluid secretion in infected ileal mucosa. Molecular Micro biology 1997;25:903 912. 6 7 Wood MW, Jones MA, Watson PR, et al. Identification of a pathogenicity island required for Salmonella enteropathogenicity. Molecular Microbiology 1998;29:883 891. 6 8 Hong KH Miller VL. Identification of a novel Salmonella i nvasion locus homologous to Shigella ipgDE Journal of Bacteriology 1998 ; 180:1793 1802. 6 9 Norris FA, Wilson MP, Wallis TS, et al. SopB, a protein required for virulence of Salmonella dublin is an inositol phosphate phosphatase. Proc Natl Acad Sci USA 19 98;95:14057 14059. 70 Eckmann L, Rudolf MT, Ptasznik A, et al. D myo inositol 1,4,5,6 tetrakisphosphate produced in human intestinal epithelial cells in response to Salmonella invasion inhibits phosphoinositide 3 kinase signaling pathways. Proc Natl Aca d Sci USA 1997;94:14456 14460. 71 Jones MA, Wood MW, Mullan PB, et al Secreted proteins of Salmonella D ublin act in concert to induce enteritis. Infection and Immunity 1998;66: 5799 5804. 72 Wood MW, Jones MA, Watson PR, et al. The secreted effector pr otein of Salmonella D ublin SopA, is translocated into eukaryotic cells and influences the induction of enteritis. Cellular Microbiology 2000;2:293 303. 73 Zhang S, Santos RL, Tsolis RM, et al. The Salmonella enterica serotype typhimurium effector prote ins SipA, SopA, SopB, SopD and SopE2 act in concert to induce diarrhea in calves. Infection and Immunity 2002;70:3843 3855. 74. Fields PI, Swanson RV, Haidaris CG et al. Mutants of Salmonella typhimurium that cannot survive within the macrophage are avir ulent, Proc Natl Acad Sci USA 1986;83:5189 5193. 7 5 Vazquez Torres A, Jones Carson J, Bumler AJ, et al. Extraintestinal dissemination of Salmonella by CD18 expressing phagocytes. Nature 1999;401:804 808. 7 6 Hoiseth SK Stocker BA. Aromatic dependent S almonella typhimurium are non virulent and effective as live vaccines. Nature 1981 ; 291 :238 9.

PAGE 187

187 7 7 Lissner CR, Swanson RN mice to Salmonella typhimurium : expression of the Ity gene in peritoneal and s plenic macrophages isolated in vitro. Journal of Immunology 1983; 131 :3006 13. 7 8 Dunlap NE, Benjamin WH, McCall Jr RD, et al Salmonella typhimurium is within splenic cells during the early phase of infection in mice. Microbial Pathogen esis 1991; 10 :297 310. 7 9 Richter Dahlfors A, Buchan AMJ Finlay BB. Murine salmonellosis studied by confocal microscopy: Salmonella typhimurium resides intracellularly inside macrophages and exerts a cytotoxic effect on phagocytes in vivo. Journal of Exp erimental Medicine 1997; 186 :569 80. 80 McSorley SJ, Asch S, Costalonga M, et al. Tracking Salmonella specific CD4 T cells in vivo reveals a local mucosal response to a disseminated Infection Immunity 2002;16:365 377. 81 Sundquist M Wick MJ. TNF alpha dependent and independent maturation of dendritic cells and recruited CD11cintCD11b1 cells during oral Salmonella infection. J Immunol 2005;175:3287 3298. 82 Salazar Gonzalez RM, Niess JH, Zammit DJ, et al. CCR6 mediated dendritic cell activation of path ogen Immunity 2006;24:623 632. 83 Formal SB. Effect of silica on the innate resistnce of inbred mice to Salmonella typhimurium infection. Infect Immun 1979;25:513 520. 84 Salcedo SP, Noursadeghi M, Cohen J et al. Intracellular replication of Salmonella typhimurium strains in specific subsets of splenic macrophages in vivo. Cell Microbiol 2001;3:587 597. 8 5 Sheppard M, Webb C, Heath F, et al Dynamics of bacterial growth and distribution within the liver during Salmonella infection. Cellular Microbiol 2003;5:593 600. 8 6 Buchmeier NA Heffron F. Inhibition of macrophage phagosome lysosome fusion by Salmonella typhimurium Infect Immun 1991;59:2232 2238. 8 7 Alpuche Aranda CM, Swanson JA, Loom is WP, et al Salmonella typhimurium activates virulence gene transcription within acidified macrophage phagosomes, ProcNatl Acad Sci USA 1992;89:10079 10083. 8 8 Rathman M, Barker LP Falkow S. The unique trafficking pattern of Salmonella typhimurium con taining phagosomes in murine macrophages is independent of the mechanism of bacterial entry. Infect Immun 1997;65:1475 1485.

PAGE 188

188 8 9 Vazquez Torres A, Xu Y, Jones Carson J, et al. Salmonella pathogenicity island 2 dependent evasion of the phagocyte NADPH Oxida se. Science 2000;287:1655 1658. 90 Uchiya K, Barbieri MA, Funato K, et al A Salmonella virulence protein that inhibits cellular trafficking. EMBO J 1999;18:3924 3933. 91 Hersh D, Monack DM, Smith MR, et al The Salmonella invasin SipB induces macropha ge apoptosis by binding to caspase 1. Proceedings of the National Academy of Sciences, USA 1999; 96:2396 2401. 92 Steneroden KK, Van Metre DC, Jackson C et al Detection and Control of a nosocomial outbreak caused by Salmonella Newport at a Large Animal Hospital. J Vet Intern Med 2010;24:606 616. 93 Holmberg S, Solomon S, Blake P. Health and economic impacts of antimicrobial resistance. Rev Infect Dis 1987 ; 9:1065 78. 94 Lee L, Puhr N, Maloney E, et al. Increase in antimicrobial resistant Salmonella infections in the United States, 1989 1990. J Infect Dis 1994 ;170:128 34. 95. Helms M, Vastrup P, Gerner Smidt P, et al Excess mortality associated with antimicrobial drug resistant Salmonella Typhimurium. Emerg Infect Dis 2002;8:490 495. 9 6 Holmberg SD, Wells JG, Cohen ML. Animal to man transmission of antimicrobial resistant Salmonella: investigations of U.S. outbreaks, 1971 1983. Science 1984 ;225:833 5. 9 7 Varma JK, Greene KD, Ovitt J, et al. Hospitalization and antimicrobial resistance in Salmonella outbreaks, 1984 2002. Emerg Infect Dis 2005 ;11:943 6. 9 8 Winokur PL, Canton R, Casellas JM, et al. Variations in the prevalence of strains expressing an e xtended spectrum b lactamase phenotype and characterization of isolates from Europe, the Americas, and the Western Pacific region. Clin Infect Dis 2001 ;32(Suppl 2):S94 103. 9 9 Molbak K. Human health consequences of antimicrobial drug resistant Salmonell a and other foodborne pathogens. Clin Infect Dis 2005;41:1613 20. 100. Travers K, Barza M. Morbidity of infections caused by antimicrobial resistant bacteria. Clin Infect Dis 2002 ; 34(Suppl 3):S131 4. 101 Carattoli A. Importance of integrons in the di ffusion of resistance. Vet Res 2001; 32 (3 4) : 243 259.

PAGE 189

189 102 Levesque C, Piche L, Larose C et al PCR mapping of integrons reveals several novel combinations of resistance genes. Antimicrob Agents Chemother 1995; 39 (1) : 185 191. 103 Carattoli A. Plasmid medi ated antimicrobial resistance in Salmonella enterica Curr Issues Mol Biol 2003; 5 (4) : 113 122. 104 Doublet B, Boyd D, Mulvey MR et al. The Salmonella genomic island 1 is an integrative mobilizable element. Mol Microbiol 2005; 55 (6):1911 1924. 10 5 Fluit A C. Towards more virulent and antibiotic resistant Salmonella ? FEMS Immunology and Medical Microbiology 2005;43:1 11. 10 6 Glenn LM, Lindsey RL, Frank JF, et al. Analysis of Antimicrobial Resistance Genes Detected in Multidrug Resistant Salmonella enteric a Serovar Typhimurium Isolated from Food Animals. Microbial Drug Resistance 2011;17:407 418. 10 7 Carattoli A, Miriagou V, Bertini A, et al. Replicon typing of plasmids encoding resistance to newer beta lactams. Emerg Infect Dis 2006;12:1145 1148. 10 8 J ohnson TJ, Wannemuehler YM, Johnson SJ, et al. Plasmid replicon typing of commensal and pathogenic Escherichia coli isolates. Appl Environ Microbiol 2007;73:1976 1983. 10 9 Lindsey RL, Fedorka Cray PJ, Frye JG, et al Inc A/C plasmids are prevalent in mul tidrug resistant Salmonella enterica isolates. Appl Environ Microbiol 2009;75:1908 1915. 1 10 Carattoli A. Resistance plasmid families in Enterobacteriaceae. Antimicrob Agents Chemother 2009;53:2227 2238. 1 11 Fricke WF, Welch TJ, McDermott PF, et al. C omparative genomics of the IncA/C multidrug resistance plasmid family. J Bacteriol 2009;191:4750 4757. 1 12 Ma D, Cook DN, Hearst JE et al Efflux pumps and drug resistance in gram negative bacteria. Trends Microbiol 1994; 2 (12);489 493. 1 13 Nikaido H. Multidrug efflux pumps of gram negative bacteria. J Bacteriol 1996; 178 (20); 5853 5859. 1 14 et al Multi drug resistance in Salmonella enterica : Efflux mechanisms and their relationships with the development of chromosomal re sistance gene clusters. Current Drug Targets 2006;7:849 860.

PAGE 190

190 11 5 Blair JMA, Ragione RML, Woodward MJ, et al. Peripalsmic adaptor protein AcrA has a distinct role in the antibiotic resistance and virulence of Salmonella enteric serovar Typhimurium. Journal of Antimicrobial Chemotherapy 2009;64:965 972. 11 6 Webber MA, Bailey AM, Blair JMA, et al. The global consequence of disruption of the AcrAB TolC efflux pump in Salmonella enterica includes reduced expression of SPI 1 and other attributes required to i nfect the host. J Bacteriol 2009; 191: 4276 4285. 11 7 Stokes HW Hall RM. A novel family of potentially mobile DNA elements encoding site specific gene integration functions: integrons. Mol Microbiol 1989; 3 (12);1669 1683. 11 8 Collis CM Hall RM. Site spe cific deletion and rearrangement of integron insert genes catalyzed by the integron DNA integrase. J Bacteriol 1992; 174 (5);1574 1585. 11 9 Chen S, Zhao S, White DG, et al. Characterization of multiple antimicrobial resistant Salmonella serovars isolated f rom retail meats. Appl Environ Microbiol 2004;70:1 7. 1 20 Peirano G, Agerso Y, Aarestrup FM, et al. Occurrence of integrons and antimicrobial resistance genes among Salmonella enterica from Brazil. J Antimicrob Chemother 2006;58:305 309. 1 2 1. von Baum H Marre R. Antimicrobial resistance of Escherichia coli and therapeutic implications. Int J Med Microbiol 2005;295:503 511. 1 22 Smith BP: Salmonella infections in horses, Compend Cont Educ Pract Vet 1981 ;3(1):S4 S13. 1 23 Hirsh DC. The alimentary canal as a microbial habitat. In: Biberstein EL, Zee YC, eds. Review of veterinary microbiology Boston: Blackwell Scientific 1990; 89 99. 1 24 Dietrich G, Griot Wenk M, Metcalfe IC, et al. Experience with registered mucosal vaccines. Vaccine 2003;21:678 683. 1 2 5 Kirby AC, Yrlid U Wick MJ. The innate immune response differs in primary and secondary Salmonella infection. J Immunol 2002;169:4450 4459. 12 6 Holt PS. Host susceptibility, resistance and immunity to Salmonella in animals. In: Wray C and Wray A, eds Salmonella in domestic animals 1st ed. London, UK: CABI Publishing, 2000;73 87.

PAGE 191

191 12 7 response to Salmonella infection as demonstrated in T cell receptor deficient mice of defined Ity genotypes. Infect Immun 1997;65:2306 2312. 128. Dunowska M, Morley PS, Traub Dargatz JL, et al. Comparison of Salmonella enterica serotype Infantis isolates from a veterinary teaching hospital. J Appl Microbiol 2007;102:1527 1536. 129. Fiala J. Salmonella outbreak shuts down New Bolton Center. DVM: The Newsmagazine of Veterinary M edicine 2004;35:6 13. 130 Rankin SC, Whichard JM, Joyce K, et al. Detection of a blaSHV Extended Spectrum {beta} Lactamase in Salmonella enterica Serovar Newport MDR AmpC J Clin Microbiol 2005;43:5792 5793. 1 31 House JK, Mainar Jaime RC, Smith BP, e t al. Risk factors for nosocomial Salmonella infection among hospitalized horses. J Am Vet Med Assoc 1999;214, 1511 1516. 1 3 2. Ekiri AB, MacKay RJ, Gaskin JM, et al. Epidemiologic analysis of nosocomial Salmonella infections in hospitalized horses. J Am V et Med Assoc 2009;234, 108 119. 1 33 Castor ML, Wooley RE, Shotts EB, et al. Characteristics of Salmonella isolated from an outbreak of equine salmonellosis in a veterinary teaching hospital. Eq Vet Sci 1989;9:236 241. 1 34 Amavisit P, Markham PF, Lightf oot D, et al. Molecular epidemiology of Salmonella Heidelberg in an equine hospital. Vet Microbiol 2001;80: 85 98. 135. Ekiri A. Unpublished data. Surveillance and Infection Control data for the Large Animal Hospital, College of Veterinary Medicine, Univer sity of Florida. Accessed 2011. 1 3 6 Morley PS. Biosecurity of veterinary practices Vet Clin Food Anim 2002;18 : 133 155. 1 3 7 Ewart SL, Schott HC, Robinson RL, et al. Identification of sources of Salmonella organisms in a veterinary teaching hospital and evaluation of the effects of disinfectants on detection of Salmonella organisms on surface materials. J Am Vet Med Assoc 2001;218:1145 1151. 1 3 8 Hyatt DR Weese JS. Salmonella culture: sampling procedures and laboratory techniques. Vet Clin N Am equine pract 2004;3:577 585.

PAGE 192

192 13 9 Funk A, Davies P R, Nichols MA. The effect of fecal sample weight on detection of Salmonella enterica in swine feces. J Vet Diagn Invest 2000; 12 : 412 418. 1 40 van Duijkeren E, Flemming C, van Oldruitenborgh Oosterbaan MS, et al Diagnosing salmonellosis in horses. Culturing of multiple versus single faecal samples. Vet Q 1995;17:63 66. 1 4 1 Ginocchio C, Rahn K, Clarke R, et al. Naturally occurring deletions in the centisome 63 pathogenicity island of environmental isolates of S almonella spp. Infect Immun 1997;65:1267 1272. 14 2 Heid CA, Stevens J, Livak KJ et al. Real time quantitative PCR. Genome Res 1996;6:986 994. 14 3 Cohen ND, Wallis DE, Neibergs HL et al. Detection of Salmonella enteritidis in equine feces using the po lymerase chain reaction and genus specific oligonucleotide primers. J Vet Diagn Invest 1995;7:219 222. 1 4 4 Cohen ND Neibergs HL McGruder ED et al Genus specific detection of salmonellae using the polymerase chain reaction (PCR). J Vet Diagn Inve st 1993 ;5(3):368 71. 14 5 Hartmann FA, Callan RJ McGuirk SM et al. Control of an outbreak of salmonellosis caused by drug resistant Salmonella anatum in horses at a veterinary hospital and measures to prevent future infections. J Am Vet Med Assoc 1996;209:629 631. 14 6 Smith BP, House JK, Magdesian K G, et al. Principles of an infectious disease control program for preventing nosocomial gastrointestinal and respiratory tract diseases in large animal veterinary hospitals. J Am Vet Med Assoc 2004;225:1186 1195. 14 7 Dunowska M, Morley PS, Patterson G, e t al. Evaluation of the efficacy of a peroxygen disinfectant filled footmat for reduction of bacterial load on footwear in a large animal hospital setting. J Am Vet Med Assoc 2006;228:1935 1939. 14 8 Stockton KA, Morley PS, Hyatt DR, et al. Evaluation of the effects of footwear hygiene protocols on nonspecific bacterial contamination of floor surfaces in an equine hospital. J Am Vet Med Assoc 2006;228:1068 1073. 14 9 Murray MJ. Salmonellosis in horses. J Am Vet Med Assoc 1996;209:558 560. 1 50 Dwyer RM Disinfecting equine facilities. Rev Sci Tech Off Int Epiz 1995;14:403 418. 1 5 1 Dwyer RM. Back to basics: disinfecting stables. The Horse 1999;40:67 80.

PAGE 193

193 1 5 2 Ernst NS, Hernandez JA, MacKay RJ, et al. Risk factors associated with fecal Salmonella shedd ing among hospitalized horse with signs of gastrointestinal tract disease J Am Vet Med Assoc 2004;225:275 281. 15 3 Morley PS. Surveillance for nosocomial infections in veterinary hospitals. Vet Clin Equine 2004;20:561 576. 15 4 Smith BP. Evolution of e quine infection control programs Vet Clin Equine 2004;20:521 530. 15 5 Traub Dargatz JL, Dargatz DA, Morley PS et al An overview of infection control strategies for equine facilities, with an emphasis on veterinary hospitals Vet Clin Equine 2004;20 : 50 7 520. 15 6 Hird DW, Casebolt DB, Carter JD, et al Risk factors for salmonellosis in hospitalized horses. J Am Vet Med Ass 1986; 188:173 177. 1 5 7 Smith B P, Reina Guerra M Hardy AJ. Prevalence and epizootiology of equine salmonellosis. J Am Vet Med Ass oc 1978;172:353 356. 15 8 Linerode PA Goode RL. The effects of colic on the microbial activity of the equine large intestine, in Proceedings 16th Annu Conv Am Assoc Equine Pract 1970;321 341. 15 9 Argenzio RA. Functions of the equine large intestine an d their interrelationship in disease. Cornell Vet 1975; 65:303 330. 1 60 Owen RA, Fullerton J Barnum DA. Effects of transportation, surgery, and antibiotic therapy in ponies infected with Salmonella. Am J Vet Res 1983; 44:46 50. 1 6 1 Clarke LL, Roberts M C Argenzio RA. Feeding and digestive problems in horses. Physiologic responses to a concentrated meal. Vet Clin North Am Equine Pract 1990; 6:433 450. 1 6 2 Qui shi L Guizhen Y. Influence of surgical stress on cellular immunity and the induction of plast ic adherent suppressor cells of spleen in mice. Immunol Invest 1986; 15:419 430. 1 6 3 Zellweger R, Ayala A, Zhu XL, et al Effect of surgical trauma on splenocyte and peritoneal macrophage immune function. J Trauma 1995; 39:645 650. 16 4 Ziegler Heitbrock HW Ulevitch RJ. CD14: cell surface receptor and differentiation marker. Immunol Today 1993; 14:121 125.

PAGE 194

194 16 5 Kawasaki T, Ogata M, Kawasaki C, et al Surgical stress induces endotoxin hyporesponsiveness and an early decrease of monocyte mCD14 and HLA DR expr ession during surgery. Anesth Analg 2001; 92:1322 1326. 16 6 Kawasaki T, Ogata M, Kawasaki C, et al. Effects of epidural anaesthesia on surgical stress induced immunosuppression during upper abdominal surgery. Br J Anaesth 2007; 98:196 203. 16 7 Traub Dar gatz JL, Salman MD, Jones RL. Epidemiologic study of salmonellae shedding in the feces of horses and potential risk factors for development of the infection in hospitalized horses. J Am Vet Med Assoc 1990;196:1617 1622. 16 8 Palmer JE, Benson CE, Whitloc k RH. Salmonella shed by horses with colic. J Am Vet Med Assoc 1985;187:256 257. 16 9 Kim LM, Morley PS, Traub Dargatz JL, et al. Factors associated with Salmonella shedding among equine colic patients at a veterinary teaching hospital. J Am Vet Med Asso c 2001;218:740 748. 1 70 Baker JR, Leyland A. Diarrhea in the horse associated with stress and tetracycline therapy. Vet Rec 1973;93:583 584. 17 1 Rhoads WS Barton MH Parks AH Comparison of medical and surgical treatment for impaction of the small colon in horses: 84 cases (1986 1996). J Am Vet Med Assoc 1999;214(7):1042 7. 17 2 Peduzzi P, Concato J, Kemper E, et al. A Simulation s tudy of the n umber of e vents per v ariable in l ogistic r egression a nalysis. J Clin Epidemiol 1996:49;1373 1379. 17 3 Owen RA, Fullerton JN, Tizard IR, et al Studies on experimental enteric salmonellosis in ponies. Can J Comp Med 1979;43:247 254. 17 4 B ryans JT, Moore BO, Crowe MW. Safety and efficacy of furoxone in the treatment of equine salmonellosis. Vet Med Small Anim Clin 1965;60:626 633. 17 5 Smith BP, Reina Guerra M, Hardy AJ, et al Equine salmonellosis: experimental production of four syndrome s. Am J Vet Res 1979;40:1072 1077. 17 6 Salmonella Anatum infection in horses. Australian Veterinary Journal 1982;58:232 240. 17 7 Sullivan A, Edlund C, Nord CE. Effect of antimicrobial agents on the ecological balanc e of human microflora. Lancet Infect Dis 2001;1:101 114.

PAGE 195

195 17 8 Huntges DJ. Resistance of the indigenous intestinal flora to the establishment of invading microbial populations. In: Schlessinger D, ed. Microbiology. Washington, DC: American Society for Micro biology, 1975;116 119. 1 7 9 Saville WJ, Hinchcliff KW, Moore BR, et al Necrotizing enterocolitis in horses: A retrospective study. J Vet Intern Med 1996;10 : 265 270. 1 80 Lavoie JP, Madigan JE, Cullor JS, et al Haemodynamic, pathological, haematological and behavioural changes during endotoxin infusion in equine neonates. Equine Veterinary Journal 1990;22:23 29. 1 8 1 McRoberts JA, Beurlein G, Dharmsathaphorn K. Cyclic AMP and Ca2+ actiavted K+ transport in a human colonic epithelial cell line. J Biol Ch em 1985;260:14163 72. 1 8 2 Duebbert IE, Peterson JW. Enterotoxin induced fluid accumulation during experimental salmonellosis and cholera: involvement of prostaglandin synthesis by intestinal cells. Toxicon 1985;23(1):157 72. 1 8 3 Waterman A. A review o f the diagnosis and treatment of fluid and electrolyte disorders in the horse. Equine Vet J 1977;9:43 48. 1 8 4 Duncan JR, Prasse KW, Mahaffey EA. In Duncan JR, Prasse KW, and Mahaffey EA (eds) Veterinary laboratory medicine: clinical pathology Erythrocyt es. 3 rd ed. Iowa State University Press, Ames. 1994 : 3 37, 249. 18 5 Tasker JB. Fluid and electrolyte studies in the horse. IV The effects of fasting and thirsting. Cornell Vet 1967;57:658 667. 18 6 Brownlow MA Hutchins DR. The concept of osmolarity: it s use in the evaluation Equine Vet J 1982;14:106 110. 1 8 7 Carter JD, Hird DW, Farver TB, et al. Salmonellosis in hospitalized horses: seasonality and case fatality rates. J Am Vet Med Assoc 1986;188:163 167. 18 8 Hanson D F. Fever, temperature, and the immune response. Ann N Y Acad Sci 1997;813:453 464. 18 9 Thurmond MC. Conceptual foundations for infectious disease surveillance. J Vet Diagn Invest 2003;15:501 514. 19 0 Windsor RC, Johnson LR, Sykes JE, et al. Molecular detection of microbes in nasal tissue of dogs with idiopathic lymphoplasmacytic rhinitis. J Vet Intern Med 2006;20:250 256.

PAGE 196

196 19 1 Baumler AJ, Tsolis RM, Ficht TA, et al Evolution of host adaptation in Salmonella enterica Infection and Immunity 1998;66:45 79 4587. 1 9 2 Collazo CM, Galan JE. The invasion associated type III protein secretion system in salmonella a review. Gene 1997;192:51 59. 1 9 3 Galan JE. Molecular genetic bases of Salmonella entry into host cells. Molecular microbiology 1996;20:263 27 1.

PAGE 197

197 BIOGRAPHICAL SKETCH Abel Ekiri received a degree in veterinary medicine from Makerere University, Kampala, Uganda in 2002 In 2006, he joined the University of Florida, Large Animal Hospital as the hospital Infection Control Officer. At the same tim e, he enrolled in graduate school and pursued Master of Science (MS) and Doctor of Philosophy (PhD) degrees with a concentration in epidemiology under the supervisi on of Dr. Jorge Hernandez. His MS and PhD research work focused on epidemiology, surveillanc e and infection control of Salmonella infections in hospitalized horses. He received his M S degree in May 2008, and will complete the PhD in May 2012.