Materials and Methods

Study design—Data were collected retrospectively for all cattle admitted to the Cornell University Equine and Farm Animal Hospital between January 1, 1996, and June 1, 2007, by use of the hospital's computerized medical records database. Variables collected for each patient included age, breed, sex, state of origin, date of admission, diagnoses, and, when performed, results of bacteriologic culture for Salmonella spp.

Detection of Salmonella organisms in the feces was conducted via 1 of 2 tests. The most commonly used test was a Salmonella surveillance culture, intended to be performed on samples obtained from all admitted cattle that did not have clinical signs consistent with salmonellosis. This test was primarily a biosecurity measure aimed at identifying subclinical shedders that posed a threat to the hospital environment. The second test was for cattle admitted with a combination of signs typical of salmonellosis (eg, diarrhea, fever, or dehydration) or for cattle originating from a farm with a history of endemic or epidemic salmonellosis (as determined on the basis of information voluntarily provided by the referring veterinarian or by recollection of the hospital clinician regarding previous admission of confirmed Salmonella-positive cattle from that farm). These patients were tested with a specific Salmonella culture or a broad enteric bacteriology culture panel that included Salmonella spp, depending on the age of the patient (calf vs adult) and preference of the attending clinician. The specific Salmonella culture was more commonly selected for samples from adult cattle, whereas the enteric bacteriology culture panel was generally used for samples from calves because Escherichia coli is a common bacterial pathogen within this age group.

Sample collection—For the purpose of Salmonella testing, clinicians and students in the veterinary medical teaching hospital were asked to routinely collect at least 10 g of fecal matter from each patient on the day of admission by use of an examination glove. The sample was packaged in a plastic specimen cup with a tight lid and then transported directly to the Cornell Animal Health Diagnostic Center for bacteriologic culture. Fecal samples collected after 5:00 PM were refrigerated overnight at 4°C prior to submission.

Microbiological procedures for Salmonella detection—Personnel at the Animal Health Diagnostic Center used standard bacteriologic culture methods to isolate Salmonella organisms from fecal samples. Salmonella culture procedures for both the surveillance and clinical suspect samples included selective enrichment as an initial step. Fecal swab specimens from each sample container were added to 10 mL of tetrathionate broth^a (10% [wt:vol]) containing 0.2 mL of iodine solution, and the mixture was incubated at 42°C for 18 to 24 hours. After incubation, the sample-broth mixture was streaked onto plates consisting of BGN^b and XLT-4 selective media, and both plates were incubated at 37°C for 18 to 24 hours. Red colonies (lactose nonfermenting bacteria) on BGN and black colonies (hydrogen sulfide–producing bacteria) on XLT-4 were inoculated into triple iron agar slants, which were then incubated at 37°C for 18 to 24 hours. The XLT-4 plates without suspected colonies were incubated at 37°C for an additional 18 to 24 hours before they were again examined for characteristic black colonies. Colonies on triple iron agar slants that had the biochemical properties of Salmonella organisms were then serogrouped by use of slide agglutination with standard protocols. Those colonies that had positive results by use of slide agglutination were subsequently identified as Salmonella spp by use of an automated microbial identification system.^c Serotyping of confirmed Salmonella isolates was performed at the USDA, APHIS National Veterinary Services Laboratories in Ames, Iowa.

In addition to the aforementioned enrichment process, direct culture of all clinical suspect samples (but not surveillance samples) was concurrently performed on solid media, regardless of whether the animal was tested with a specific Salmonella culture or an enteric bacteriology culture panel that included Salmonella spp. This was intended to enable clinicians to more rapidly reach a diagnosis when they were suspicious of Salmonella infection. Fecal swab specimens from each sample container were directly inoculated onto trypticase soy blood agar with 5% sheep blood, Levine eosin methylene blue agar, and BGN.

Statistical analysis—Data were imported into a commercially available statistical software program^d for variable coding and analysis. Age was converted into a categoric variable (calf, < 1 year old; yearling, ≥ 1 year old but < 2 years old; and adult, ≥ 2 years old). Date of admission was used to create a variable for season (winter, December through February; spring, March through May; summer, June through August; and fall, September through November). Breed and state of origin were converted into dichotomous variables (Holstein vs other and New York vs other, respectively). Several diseases were chosen as diagnosis predictor variables because they represented the bulk of the bovine inpatient caseload at our veterinary medical teaching hospital: ruminal bloat, LDA, RDA (with or without volvulus), traumatic reticuloperitonitis (ie, hardware disease), enteritis, cecal dilatation-volvulus, fatty liver syndrome, pneumonia, lymphosarcoma, mastitis, metritis, dystocia, umbilical hernia, septicemia, and diseases causing lameness. Diagnoses with an obvious age restriction (such as dystocia and umbilical hernia) were limited to models involving the relevant age group.

The prevalence of fecal shedding of Salmonella organisms among tested bovine patients was calculated. Cattle tested for Salmonella spp > 1 time were considered positive when at least 1 bacteriologic culture of a fecal sample yielded a positive result. Univariable descriptive analysis was performed on all predictor variables. Bivariable analysis via the F² test was used to determine whether each variable was significantly associated with Salmonella status (positive vs negative). Multivariable logistic regression models were used to identify possible risk factors for fecal shedding of Salmonella spp among hospitalized cattle, with Salmonella status used as the dichotomous outcome variable. Separate logistic regression models were used for each age group because some of the risk factors investigated were unique to particular age groups. Initial selection of variables was based on the bivariable analysis screening (P < 0.25), and a backward stepwise approach was used to identify a final multivariable model for each age group; values of P < 0.05 were considered significant for the final model. Relevant 2-way interaction terms were also investigated for significance within each model. Serotype information on all cattle with positive test results was obtained from the medical records.

Results

During the study period, 5,398 cattle were admitted to our veterinary medical teaching hospital. Of these, 1,438 (26.6%) were calves, 303 (5.6%) were yearlings, and 2,708 (50.2%) were adult cattle; age was not recorded for 949 (17.6%) cattle. A total of 4,790 (88.7%) were females, 4,775 (88.5%) were Holsteins, and 4,364 (80.8%) were from the state of New York. Caseload was highest in the spring (1,565 [29.0%] cattle) and lowest in the fall (1,139 [21.1%]); caseload for summer and winter was 1,396 (25.9%) and 1,298 (24.0%) cattle, respectively. Disease distribution was as follows: LDA (777 [14.4%] cattle), pneumonia (598 [11.1%]), enteritis (486 [9.0%]), RDA (423 [7.8%]), metritis (356 [6.6%]), mastitis (304 [5.6%]), dystocia (190 [3.5%]), fatty liver syndrome (150 [2.8%]), septicemia (138 [2.6%]), umbilical hernia (137 [2.5%]), cecal dilatation-volvulus (98 [1.8%]), traumatic reticuloperitonitis (82 [1.5%]), ruminal bloat (79 [1.5%]), lameness (65 [1.2%]), and lymphosarcoma (44 [0.8%]). Among cattle tested with the Salmonella surveillance culture, 66 (17.6%) calves and 177 (14.9%) adults had fecal samples collected 2 or more times for bacteriologic culture; among cattle tested as clinical salmonellosis suspects, 109 (36.8%) calves and 69 (20.4%) adults had fecal samples collected 2 or more times for bacteriologic culture. Because of the relatively small number of yearling cattle in the study, these animals were excluded from the statistical analysis.

Among all cattle, 2,020 (37.4%) were tested via the Salmonella surveillance culture, and the prevalence of fecal shedding of Salmonella organisms was 2.5% (50/2,020) within this group. Of the 374 calves tested via the Salmonella surveillance culture, 12 (3.2%) had positive results, whereas 27 of 1,185 (2.3%) adult cattle in this group had positive results (Table 1). Male cattle, cattle of non-Holstein breeds, and cattle admitted during the summer were more likely to have positive culture results. Among the disease categories, the probability of having a positive Salmonella culture result ranged from 0% for lymphosarcoma or lameness to 8.8% for septicemia. However, bivariable analysis via χ² testing failed to detect a significant relationship between any of these variables and Salmonella surveillance culture status.

A total of 768 of the 5,398 (14.2%) cattle were classified as clinical salmonellosis suspects and tested via a specific Salmonella culture or an enteric bacteriology culture panel that included Salmonella spp; the prevalence of fecal shedding of Salmonella was 6.5% (50/768) within this group. Bivariable analysis revealed that the prevalence of fecal shedding of Salmonella spp was significantly higher (P = 0.004) in calves (27/296 [9.1%]) than in adult cattle (12/338 [3.6%]; Table 2). There was also a significant (P < 0.001) difference in the seasonal distribution of cattle that had positive culture results, with the highest proportion in the fall (23/165 [13.9%]) and the lowest in the spring (7/214 [3.3%]). The prevalence of Salmonella shedding was significantly (P = 0.01) higher in male cattle (10/74 [13.5%]) than in female cattle (40/694 [5.8%]). Breed and state of origin were not significantly associated with Salmonella shedding status among the clinical suspect cattle. The only diseases significantly associated with Salmonella shedding status among clinical suspects were enteritis (P < 0.001) and septicemia (P = 0.003).

Risk factors for Salmonella shedding—Multivariable logistic regression models were developed for calves and adult cattle tested via the Salmonella surveillance culture. It was found that none of the study variables were significant predictors of Salmonella shedding status.

Multivariable logistic regression analysis for clinical suspect cattle was also performed via separate models for each age group. Calves admitted during the fall were significantly (P = 0.004) more likely to be shedding Salmonella organisms than were calves admitted in the spring (OR, 5.9; 95% CI, 1.8 to 19.6; Table 3). Among the disease categories, calves with septicemia (OR, 3.3; 95% CI, 1.1 to 10.4; P = 0.04) or an umbilical hernia (OR, 8.6; 95% CI, 1.3 to 56.8; P = 0.03) were significantly more likely to be shedding Salmonella organisms than were calves without those diseases. Adult cattle admitted during the fall had a nonsignificant (P = 0.15) increase in the likelihood of Salmonella shedding (OR, 3.7; 95% CI, 0.6 to 22.5; Table 4), compared with the likelihood of Salmonella shedding among adult cattle admitted in the spring. Regarding the disease categories, adult cattle with enteritis (OR, 9.9; 95% CI, 2.9 to 34.6; P < 0.001) or metritis (OR, 5.2; 95% CI, 1.4 to 19.1; P = 0.01) were more likely to be shedding Salmonella organisms than were adult cattle without those diseases.

Salmonella serotypes—We identified 21 Salmonella serotypes during the study period (Table 5), with the most common being S enterica serotype Typhimurium, including the Copenhagen variant (33/100 [33%] isolates), S enterica serotype Newport (23 [23%]), and S enterica serotype Agona (12 [12%]). Sixteen serotypes were represented among the cattle testing positive via the surveillance culture, with Salmonella Typhimurium and Salmonella Newport each accounting for 12 of 50 (24%) isolates. Twelve serotypes were detected among the clinical suspects with positive culture results, with the predominant serotype being Salmonella Typhimurium (21/50 [42%] isolates). Among all cattle with positive culture results, a clear temporal pattern was evident for the 2 most common serotypes: 30 of 33 (91%) Salmonella Typhimurium isolates were obtained from cattle admitted between 1996 and late 2001, whereas all of the Salmonella Newport isolates were obtained from cattle admitted between 2002 and 2007.

Discussion

The prevalence or risk factors (or both) for fecal shedding of Salmonella spp among hospitalized horses have been described in several studies that have focused on the general population of equine inpatients^5,6,8–10 or horses specifically examined because of gastrointestinal tract disease.^4,7 In contrast, few studies have been published on hospitalized cattle. The study reported here had the advantage of the use of data in medical records gathered on bovine patients over a period of > 11 years, and it examined both the broad population of admitted cattle as well as cattle admitted with a combination of clinical signs compatible with a diagnosis of salmonellosis (eg, diarrhea, fever, and dehydration). Data from 5,398 cattle examined at our veterinary medical teaching hospital were available for the study, although fecal samples from slightly more than half (2,788 [51.6%]) of these cattle were tested via some form of Salmonella culture. Bacteriologic culture of feces does not have perfect sensitivity for detecting Salmonella shedding, and we recognize that some cattle that were shedding Salmonella organisms were presumably missed by culturing and thus misclassified during the study. However, negative culture results in a specific animal would suggest that substantial fecal shedding of Salmonella organisms and hospital contamination were less likely. Furthermore, 37% and 20% of calves and adult cattle that were tested as clinical suspects, respectively, were cultured for fecal shedding of Salmonella spp multiple times during hospitalization.

Among the general population of cattle not suspected of having salmonellosis, the prevalence of fecal shedding of Salmonella spp was 2.5%. Those cattle with clinical signs typical of salmonellosis had a 6.5% prevalence of Salmonella shedding. The prevalence of 2.5% among the general population is comparable to results obtained by researchers in California (3.5% during a 10-year period)¹¹ and Montreal (1.4% during a 1-year period)¹²; in both of those studies, investigators estimated the prevalence of Salmonella shedding among hospitalized cattle irrespective of clinical signs at the time of admission. Estimates of prevalence for fecal shedding of Salmonella spp among the overall population of hospitalized horses have ranged from 3% to 7%,^6,8,13 whereas estimates for horses examined specifically because of diseases of the gastrointestinal tract have ranged from 9% to 13%.^4,7,14 Our estimates of the prevalence of Salmonella shedding among hospitalized cattle may be lower than those that would be obtained at other veterinary teaching hospitals because of the relatively low prevalence of Salmonella spp among cattle in the state of New York. The prevalence of fecal shedding of Salmonella spp among cattle in central New York has been estimated at 1.7%,¹⁵ and > 80% of the bovine caseload at our veterinary medical teaching hospital during the study period originated from the state of New York. According to the 2002 USDA National Animal Health Monitoring System report,¹⁶ the herd-level prevalence of Salmonella shedding was significantly lower in the Northeast than in all other geographic regions of the country.

Age group was significantly associated with prevalence of Salmonella shedding among those cattle admitted to the hospital as clinical suspects. Prevalence of shedding was > 9% in calves and 3.6% in adult cattle. Among nonsuspect animals tested via the Salmonella surveillance culture, the prevalence was also higher in calves than in adult cattle, but this was not a significant difference. Foals with gastrointestinal tract disease were more likely to shed Salmonella organisms than were adult horses with gastrointestinal tract disease in 1 study,⁴ but other studies^5,6 have failed to detect an association between age and shedding among horses. Both humoral and cellular immune mechanisms play a role in resistance to Salmonella organisms,¹ and calves may be at a greater risk of infection than adult cattle because of their more naïve immune system. In addition, concurrent infection with multiple enteric pathogens is a common scenario in calves, further compromising their immune status.¹ It is also possible that a relative lack of intestinal microflora in young calves contributes to their susceptibility; such microflora is believed to offer a degree of protection against colonization by pathogenic enteric bacteria.

Season was also significantly associated with prevalence of Salmonella shedding among those cattle admitted to the hospital as clinical suspects. Prevalence was highest in the fall (approx 14%) and lowest in the spring (approx 3%). Among nonsuspect cattle tested via the Salmonella surveillance culture, prevalence was highest in the summer and lowest in the spring, but these differences were not significant. These results are in agreement with those of 2 studies^10,17 on hospitalized horses in which investigators evaluated the seasonality of Salmonella shedding. In one of those studies,¹⁰ investigators found that the number of cases reached a peak in September and a low point in March, on the basis of data gathered over a period of 11 years. In the other study,¹⁷ investigators found the prevalence of shedding during a 1-year period to be highest between late summer and early fall, whereas it was lowest in the spring. It has been reported in 1 study¹⁵ that fecal shedding of Salmonella organisms among dairy cattle on farms is more common in the summer and fall, and investigators in other studies^18–20 have detected an increase in shedding during the summer months on dairy farms and at livestock markets. This seasonal association may be related to temperature or moisture conditions that prevail in the summer and fall months; an increase in mean daily ambient temperature has been reported to be a risk factor for fecal shedding of Salmonella spp among hospitalized horses.⁸ It is possible that the ability of Salmonella spp to thrive and persist in warm, moist environments makes it more likely that various host species will come in contact with this organism and become infected. The effect of heat stress on cattle is another potential explanation because such physiologic stress could predispose cattle to intestinal colonization by Salmonella spp and other pathogenic bacteria. However, cooling cattle by use of water sprinkling does not have a significant effect on the incidence of fecal shedding of Salmonella spp.²¹ Interestingly, the peak incidence of salmonellosis in humans has been reported by numerous sources to be during the summer months,^22–25 although this is often blamed on factors such as hygiene issues associated with outdoor cooking, increased recreational water use, and a greater tendency to travel abroad.

A number of diseases were examined to determine whether they were associated with an increased likelihood of Salmonella shedding among hospitalized cattle. These particular diseases were selected because they represented most of the conditions affecting cattle admitted to our veterinary medical teaching hospital. Cattle with primary gastrointestinal tract disease (other than enteritis) during the study period were more likely to be tested via the Salmonella surveillance culture than were cattle with diseases of other body systems (Table 1). This may have been attributable to the fact that per rectal examinations would have been performed on these cattle during the initial physical examination, making it more likely that a surveillance sample was submitted. However, none of the disease variables were significant predictors of Salmonella shedding status, even when separate models were used for each age group. Thus, an animal admitted because of dystocia or mastitis, for example, was just as likely to have positive results for the Salmonella surveillance culture as an animal admitted because of an LDA or traumatic reticuloperitonitis. Even a diagnosis of enteritis among cattle tested via surveillance culturing (rather than as clinical suspects) was not significantly associated with Salmonella status; this was likely related to the mild clinical signs they had, such that they were not considered salmonellosis suspects. These results suggest that among the general population of bovine patients not suspected of having salmonellosis, targeted screening of cattle with some form of primary gastrointestinal tract disease is not warranted. Rather, all such cattle should be tested with equal vigor. These results concur with those of a study⁶ of an overall population of hospitalized horses in which investigators failed to detect a significant association between Salmonella shedding and reason for admission. Similarly, investigators of another study⁹ involving a general population of equine patients concluded that admission because of gastrointestinal tract disease was not significantly associated with Salmonella status; in fact, such horses were less likely to have positive culture results than were those admitted because of respiratory tract or neurologic disease. In contrast, admission because of colic was found in 1 study⁵ to be a significant risk factor for positive Salmonella status among a general population of hospitalized horses.

Cattle with typical signs of salmonellosis (or originating from a farm with a known history of salmonellosis) were tested via a specific Salmonella culture or an enteric bacteriology culture panel that included Salmonella spp, as opposed to the surveillance culture performed for biosecurity reasons. Among calves, a diagnosis of septicemia or an umbilical hernia was associated with a greater likelihood of shedding Salmonella spp in the feces. The diagnosis of septicemia is used by clinicians at our veterinary medical teaching hospital to indicate presumed or confirmed disease in multiple organ systems associated with pathogenic bacteria or their toxins in the bloodstream. Because septicemia is common in calves with salmonellosis, an association with Salmonella shedding was not unexpected. This propensity for sepsis, combined with the dehydration and electrolyte disorders that accompany profound fluid loss, is the reason that the mortality rate attributable to Salmonella infections is greater in calves than in adult cattle.¹ In fact, salmonellosis in calves may result in a peracute septicemia that causes death before diarrhea even becomes evident.¹ The relationship between Salmonella shedding and a diagnosis of umbilical hernia is less clear. Hernias among calves in our veterinary hospital are often complicated by an infectious process, either a cellulitis-abscess or an umbilical remnant infection. Perhaps the concomitant infection has an immunosuppressive effect on these calves, thus increasing their susceptibility to Salmonella colonization. A diagnosis of enteritis was not associated with a greater likelihood of shedding Salmonella organisms among calves in the study reported here. This was not a surprising finding because salmonellosis is a relatively uncommon diagnosis in young calves admitted to our hospital with diarrhea. More commonly, calves admitted because of enteritis are infected with E coli, rotavirus, coronavirus, Cryptosporidium spp, or a combination of these pathogens.

Among adult cattle, a diagnosis of enteritis or metritis was associated with increased odds of shedding Salmonella spp. Enteritis is a hallmark finding in adult cattle with salmonellosis, so an association with Salmonella shedding was anticipated. Diarrhea in these patients varies in severity, and the feces may contain blood and mucus; the accompanying foul smell has been described as a septic tank odor.¹ This clinical syndrome, particularly in conjunction with a fever, should always raise the index of suspicion for a diagnosis of salmonellosis in cattle. The relationship between Salmonella shedding and a diagnosis of metritis is not apparent. Metritis results from a bacterial infection of the uterus following parturition, which can lead to systemic signs of toxemia in some cows. It is possible that this infectious process exerts an immunosuppressive effect. Furthermore, these cattle may have some degree of gastrointestinal stasis as a result of their primary disease, and any subsequent alterations in intestinal microflora could offer an advantage to pathogenic enteric bacteria. Metritis also serves as a marker for the postpartum period, which is a time of substantial physiologic stress for dairy cows. Perhaps the combination of postpartum stress and concurrent infection, along with any gastrointestinal stasis that may exist, enhances the susceptibility of these cattle to Salmonella colonization. In contrast to our results in calves, a diagnosis of septicemia was not associated with a greater likelihood of Salmonella shedding among adult cattle. This is a predictable result because adult cattle with salmonellosis are less likely than calves to develop bacteremia,²⁶ and clinical evidence of multiple organ system involvement is not typical among adult cattle infected with Salmonella organisms.¹

The most commonly detected serotypes in our study were Salmonella Typhimurium, Salmonella Newport, and Salmonella Agona, which accounted for 68% of the isolations. Two studies^11,12 of Salmonella shedding among hospitalized cattle also revealed that Salmonella Typhimurium was the most frequently isolated serotype. In addition, these findings are consistent with recent data from the CDC that indicated the 3 most common serotypes isolated from clinical bovine samples were Salmonella Typhimurium, Salmonella Newport, and Salmonella Agona, followed by S enterica serotype Dublin and S enterica serotype Montevideo.²⁷ In the study reported here, Salmonella Newport and Salmonella Agona isolations were quite evenly distributed between the cattle tested via the surveillance culture and the clinical suspects, but the Copenhagen variant of Salmonella Typhimurium was overrepresented within the clinical suspect cattle. This suggests that this Salmonella Typhimurium variant may be especially pathogenic in cattle and thus more likely to cause clinical disease, as opposed to being found via routine surveillance in cattle that do not have clinical signs of salmonellosis. The relative dearth of Salmonella Dublin isolations in our study reflects the fact that this serotype was not common among dairy farms in the northeastern United States during the study period.¹

The clear temporal demarcation for the 2 most common serotypes was particularly striking because > 90% of the cattle with Salmonella Typhimurium were admitted during the first half of the study period, whereas all of the cattle with Salmonella Newport were admitted during the second half. This temporal shift reflects the potential for rapid and widespread emergence of Salmonella serotypes, and it seems to coincide with the historical incidence of Salmonella infections in people. During the early 1990s, MDR Salmonella Typhimurium definitive phage type 104 emerged across the United States.²⁸ By 1998, 30% of all foodborne Salmonella infections in the United States were caused by Salmonella Typhimurium.²⁹ In 2000, however, the CDC detected a sharp increase in the incidence of infections attributable to Salmonella Newport, primarily as a result of MDR strains.³⁰ Salmonella Newport has continued to increase in importance as a pathogen for humans, whereas Salmonella Typhimurium appears to be less prevalent. According to CDC FoodNet data from 2006,³¹ the annual incidence of foodborne infections attributable to Salmonella Newport in the United States had increased by 42% over the average annual incidence for 1996 through 1998. Meanwhile, the annual incidence of foodborne infections attributable to Salmonella Typhimurium had decreased by 41%, compared with the average annual incidence for the same baseline period.³¹ The predominance of Salmonella Newport among hospitalized cattle in our study between 2002 and 2007 is noteworthy because dairy cattle are considered an important reservoir for MDR Salmonella Newport infections in people.^30,32,33