Materials and Methods

Animals—The 2 herds enrolled in this study were identified during a previous prospective study¹⁷ of > 800 dairy herds conducted to determine the incidence of clinical salmonellosis in dairy cattle in the northeastern United States. Herds for that incidence study were recruited by herd veterinarians who were asked to include all willing clients with ≥ 30 dairy cattle. The study included the option for participants to also take part in a related case-control investigation and in the fecal shedding study reported here. To be included in the present study, farms were required to meet the following criteria: initially negative for salmonellosis based on monitoring during the incidence study; clinical Salmonella infection diagnosed in preweaned calves during the incidence study and confirmed with diagnostic laboratory testing; within a 4-hour drive from Ithaca, NY; and herd owner or manager consent obtained. The protocol for the present study was approved by the Animal Use and Care Committee of Cornell University.

Herd A, the first to be enrolled, was a free-stall Holstein dairy herd of 300 cows. The farm had no history of vaccinating cows against Salmonella spp, but nonlactating cows were vaccinated with a commercially available vaccine against the agents that cause coliform mastitis.^a In the first week of February 2005, after confirmation of the outbreak described here and approximately at the time the present study began in this herd, the farm staff started vaccinating late-gestation pregnant heifers, cows near calving, and lactating cows with a vaccine^b for the control of disease and fecal shedding associated with Salmonella Newport infection. A booster vaccine was administered 2 to 3 weeks after initial vaccination.

This herd had 30 to 40 calves < 6 weeks of age. At least 4 L of colostrum was provided to each calf via bottle or tube feeder within the first 24 hours after birth. Afterward, calves were fed whole milk from vaccinated and unvaccinated cows after the milk was pasteurized and a coccidiostat (decoquinate) was added. When whole milk was not available (which was uncommon), calves were fed milk replacer containing a coccidiostat. All calves also had access to a concentrate feed that also contained a coccidiostat. All calves were housed in a greenhouse facility with separate wire pens that abutted each other. These pens were bedded with sand topped with sawdust in the summer and straw in the winter. Feces and soiled bedding were removed daily, and fresh sawdust or straw was added as necessary. Amprolium was used as needed for treatment of coccidiosis. Calves were typically weaned at 4 months of age.

Ceftiofur, penicillin, or tilmicosin was typically used to treat calves and weaned heifers with respiratory disease. Before the study, ceftiofur was typically used to treat calves with neonatal diarrhea.

Herd B, enrolled later, included 400 cows housed in a free-stall barn. From May 2004 through April 2005, the herd was expanded through the purchase of approximately 40 pregnant heifers that were close to calving and 60 weaned heifers raised within the herd. Nonlactating cows were vaccinated with 1 of 2 commercially available vaccines against the agents that cause coliform mastitis.^a,c Within 1 month after initial detection of Salmonella shedding (in June 2005), all pregnant heifers close to calving and nonlactating cows were vaccinated against Salmonella Newport.^b

Herd B had approximately 30 to 50 calves at any given time. Between 2 and 4 quarts of colostrum was provided to calves via bucket or bottle within the first 24 hours after birth. Calf buckets were not washed or disinfected on a regular basis but were washed with water about 5 times/y. Calves were fed a 23% protein, 18% fat milk replacer and a concentrate feed. The diet was routinely supplemented with a coccidiostat (decoquinate) and ionophores (lasalocid sodium or monensin). Calves were typically weaned at 6 weeks of age, but some were weaned at 5 weeks of age. Housing for the calves was similar to that of farm A.

All calves < 1 month of age were treated with TMS, beginning within 48 hours after birth. Calves that appeared healthy were treated once per day for 2 weeks, and sick calves were treated twice per day for 3 weeks. No antimicrobials were routinely used in feed or water in weaned calves or heifers. Ceftiofur, florfenicol, and tilmicosin were used to treat calves and heifers with respiratory disease. Florfenicol was typically used for systemic treatment of calves with neonatal diarrhea; however, within 3 weeks after the salmonellosis outbreak began, the antimicrobial was changed to TMS because florfenicol did not appear to be effective. When this change in antimicrobials took place, treatment of all calves between 2 to 3 weeks of age with TMS was initiated as a preventive measure.

Sample collection—In each herd, samples from calves suspected of having salmonellosis and other calves were collected within 1 to 3 weeks after recognition of cattle with clinical salmonellosis to obtain an initial estimate of the prevalence of fecal Salmonella shedding. Thereafter, samples included in this study did not include samples collected for diagnostic testing as part of routine veterinary service and were collected on a regular schedule from all enrolled calves without regard to clinical illness. Herd A was more accessible for sample collection; therefore, fecal samples were collected from this herd twice per week. Because herd B was farther away and because preliminary results from the first herd suggested that less frequent sampling would be adequate to meet study objectives, samples were collected from herd B once per week. Results of bacteriologic culture and antimicrobial susceptibility testing were shared with owners and veterinarians, and the 2 herds continued to receive clinical service and management advice related to salmonellosis from their regular veterinarians during the study.

For serologic testing, approximately 3 mL of blood was collected from a jugular vein of each calf into a sterile 5-mL glass collection tube. Whole blood samples were stored on ice until serum could be separated via centrifugation (within 24 hours after sample collection). Approximately 1 mL of serum was transferred to a plastic vial and stored at −20°C until shipment to the Danish Institute for Food and Veterinary Research for serologic testing.

Once a calf herd was identified as positive for Salmonella infection and enrolled in the present study, swabs were used to obtain fecal samples to decrease the time required for sample collection. A swab specimen from the rectum of each calf was obtained and placed into a plastic transport device containing Amies transport medium with charcoal.^d Samples were then placed in insulated boxes containing ice packs and taken directly or shipped overnight to the Animal Health Diagnostic Center at Cornell University. Samples were collected from each enrolled calf until weaning, death, or conclusion of the study.

Isolation of Salmonella—Standard culture methods¹⁸ were used to recover Salmonella spp from fecal specimens. Bacteriologic culture procedures included enrichment of sample material in tetrathionate broth,^e selection of potential Salmonella isolates by use of brilliant green novobiocin agar^f and xylose-lysine-tergitol-4 agar, and biochemical identification of suspect isolates with Kligler iron agar slants^f and an automated identification system.^g Salmonella isolates were identified by somatic serogroup classification (B, C1, C2, D1, and E) by means of a slide agglutination test. Confirmed Salmonella isolates were sent to the National Veterinary Services Laboratory of the USDA for serotyping. On the same day of biochemical confirmation of Salmonella spp, cultures were stored at room temperature (approx 22°C) on TSA slants and were lyophilized in skim milk.

Antimicrobial susceptibility testing—A broth microdilution method was used to determine the MICs of all isolates with a panel of 15 antimicrobials: amikacin, amoxicillin-clavulanic acid, ampicillin, cefoxitin, ceftiofur, ceftriaxone, chloramphenicol, ciprofloxacin, gentamicin, kanamycin, nalidixic acid, streptomycin, sulfisoxazole, tetracycline, and TMS. Most isolates were tested for antimicrobial susceptibility within a week after isolation; however, a few isolates (n = 9) were recovered for MIC testing after approximately 1 year of storage. Isolates stored on TSA slants were recovered for antimicrobial susceptibility testing by overnight subculture on TSA with 5% sheep blood. Lyophilized isolates were recovered for antimicrobial susceptibility testing by reconstitution in water, followed by overnight subculture on 5% sheep blood agar. The MICs of Salmonella isolates were determined by use of a semiautomated antimicrobial susceptibility testing system,^h following the manufacturer's instructions. For each antimicrobial, the minimum dilution that inhibited growth of the Salmonella isolate was considered the MIC. Quality control was performed every week that antimicrobial susceptibility testing was conducted by use of the following 4 bacteria from the American Type Culture Collection: Escherichia coli ATCC 25022, Staphylococcus aureus ATCC 29213, Enterococcus faecalis ATCC 29212, and Pseudomonas aeruginosa ATCC 27853. The CLSI-recommended MIC ranges for quality control^19,20 were used when available, and results were consistently within these ranges.

The CLSI interpretive criteria were used to classify Salmonella isolates as resistant or not resistant to antimicrobials on the basis of results for MICs.^19,20 The CLSI resistance breakpoints were based on human data for Enterobacteriaceae. No interpretive criteria for Enterobacteriaceae were available for ceftiofur or streptomycin, so the resistant breakpoints reported by the National Antimicrobial Resistant Monitoring System in 2000 were used for these antimicrobials.²¹

PFGE pattern determination—Pulsed-field gel electrophoresis was performed in accordance with the CDC PulseNet protocol.²² The restriction enzyme XbaI (50 U/sample) was used to digest chromosomal DNA, and PFGE was performed with a PFGE system.ⁱ Electrophoresis conditions included an initial switch time of 2.16 seconds, a final switch time of 63.80 seconds, and a run time of 21 hours. The CDC Salmonella serovar Branderup H19812 was used as the reference strain. Pictures of PFGE gels were obtained with a molecular imager.^j Comparison analysis was performed with commercial software.^k Similarity analysis was performed by calculation of the Dice coefficient, and clustering was created by use of the unweighted pair group method with arithmetic mean.

Serologic testing—The method for serologic testing for Salmonella infection is described elsewhere, with slight modifications.^23–26 Briefly, serum was tested for IgG against Salmonella Typhimurium with a mixed ELISA based on LPS O antigens 1, 4, 5, and 12 from Salmonella Typhimurium and O antigens 1, 9, and 12 from Salmonella Dublin. Percentage OD was calculated by comparing the sample OD to results for positive and negative reference samples. A cutoff OD of 25% was used to classify cattle as seropositive as recommended by the testing laboratory. Although this assay also detects IgG against Salmonella Dublin, no Salmonella Dublin infections had been identified in the herds used in our research projects.

Statistical analysis—To determine whether the percentage of Salmonella isolates with resistance to ceftriaxone increased with time in herd A, a logistic-transformed proportional hazards model was used.^l In this model, the ceftriaxone MIC for each isolate was included as the response variable and days from the date the index case (ie, first laboratory-confirmed case of Salmonella shedding) was detected were included as a fixed-effect explanatory variable. The fixed-effect partial-likelihood test method was used to produce a conditional (subject-specific) estimate while controlling for the repeated sampling. Because all isolates were susceptible to at least the highest concentration tested, none of the isolate data were excluded from the analysis.

The age at which calves were enrolled was defined as the calf age when the earliest fecal sample was collected. The Kaplan-Meier life test method was used to estimate the duration of Salmonella fecal shedding among calves.^m For each calf, the duration of shedding was defined as the interval from the date of the first Salmonella-positive fecal sample to the date of the last Salmonella-positive fecal sample collected. When a calf had only 1 positive test result for Salmonella spp, the duration of shedding was recorded as 1. Calves without 3 consecutive negative test results after the last positive test result were considered to be right-censored. Three consecutive negative test results could not be obtained for some calves because of death or removal from the calf group. Because fecal samples were not obtained from calves every day, the exact start and end date of fecal shedding was unknown. Therefore, duration of shedding was estimated as half of the interval between the date of the positive test result and the date of the prior or subsequent negative test result (or sample collection date when a calf was lost to follow-up). For these calculations, it was assumed that calves could not have been infected with Salmonella spp before the index case of shedding in the same herd, even if those calves were born before the index case. Analyses of shedding duration were performed separately for actual and estimated shedding times.

The association between fecal shedding status and seroconversion indicating Salmonella infection (from seronegative to seropositive; OD > 25%) was determined by use of the Fisher exact test.ⁿ Only animals that were seronegative (ie, OD ≤ 25%) when the first fecal sample was collected were included in this analysis, and calves that were seropositive on subsequent visits were classified as having seroconverted. A calf was classified as having a positive fecal test result when Salmonella spp was isolated from at least 1 fecal sample over the entire study period. In addition, the agreement between fecal shedding status and seroconversion was measured by calculation of the kappa statistic.^o A kappa value between 1 and 0.76 was used to define excellent agreement beyond chance between shedding status and seroconversion, a value between 0.75 and 0.41 was used to define good to fair agreement, and a value between 0.40 and 0 was used to define poor agreement beyond chance.²⁷ In addition, the association between the maximum serum anti-Salmonella IgG titer measured during the study period and fecal shedding of Salmonella spp was determined by use of the Savage score test (a linear rank test statistic) because this test can be used to evaluate scale shifts in exponential distributions.^p

An association was also evaluated between high serum IgG titer within the first week after birth and fecal shedding of Salmonella spp during the study period. This analysis was limited to calves from herd B from which samples had been collected within 1 week after birth and from which samples continued to be collected for at least 4 weeks thereafter. The Fisher exact test was used to compare the proportion of calves that were seropositive within the first week after birth and had a positive fecal test result at least once during the study period with the proportion of calves that were seronegative within the first week after birth and had a positive fecal test result at least once during the study period. We also examined the association between the serologic OD value for IgG against Salmonella Typhimurium within the first week after birth and subsequent fecal shedding of Salmonella spp by use of the Savage score test.ⁿ Values of P < 0.05 were considered significant for all analyses.

Results

Animals—In herd A, 20 calves were enrolled; the median age at enrollment was 3.5 days (range, 1 to 40 days). In herd B, 79 calves were enrolled; the median age at enrollment was 5 days (range, 1 to 35 days). Although clinical signs of salmonellosis were not formally recorded as part of the study, most of the tested calves appeared healthy. There was no obvious association between clinical illness and results of bacteriologic culture.

Sample collection—In herd A, twice-weekly collection of fecal samples and weekly collection of blood samples began on February 7, 2005 (26 days after detection of the laboratory-confirmed index case of Salmonella shedding). On this date, the 12 youngest calves were enrolled, and thereafter, calves between 1 and 4 days of age were enrolled until a total of 20 calves was obtained. Among the 12 calves enrolled on February 7, 2005, a fecal sample also had been tested for Salmonella from 2 and 8 calves on January 25, 2005, and January 30, 2005, respectively. Each calf enrolled in the study was monitored until weaning, death, or the end of the study (March 29, 2005), whichever happened earlier. The shortest follow-up time was 4 weeks.

The frequent testing program for preweaned calves began on June 20, 2005, 13 days after the laboratoryconfirmed index case was identified. We enrolled the 30 youngest calves on the farm at that time and thereafter enrolled calves between birth and 7 days of age. Blood and fecal samples were collected once per week for 12 weeks. For calves in which fecal shedding of Salmonella was never detected, blood and feces were collected once a week until weaning or until September 4, 2005, whichever came first. For calves in which fecal shedding of Salmonella spp was detected, we continued to collect fecal samples until 3 consecutive negative fecal-test results were obtained or until September 4, 2005, whichever came first.

Salmonellosis history—Prior to December 13, 2004, Salmonella culture results for all fecal and environmental samples collected from herd A as part of other studies were negative. A fecal sample collected on January 12, 2005, from a clinically ill calf tested positive for Salmonella Typhimurium variant Copenhagen. This calf represented the laboratory-confirmed index case of Salmonella shedding. Subsequently, fecal samples from 6 clinically ill cows and 4 clinically ill calves collected on January 25, 2005, tested positive for Salmonella spp. All cow and calf samples were serotyped as Salmonella Typhimurium variant Copenhagen. Results of antimicrobial susceptibility testing of the Salmonella isolates recovered from all 5 calves revealed similar susceptibility patterns. Three isolates were resistant to 9 antimicrobials (amoxicillin-clavulanic acid, ampicillin, cefoxitin, ceftiofur, chloramphenicol, kanamycin, streptomycin, sulfisoxazole, and tetracycline), and 2 isolates were resistant to 10 antimicrobials (the same 9 antimicrobials plus ceftriaxone). In addition, Salmonella Typhimurium variant Copenhagen was recovered from samples collected on January 30, 2005, from 3 apparently healthy calves. One isolate had the same 9-antimicrobial resistance pattern as the 3 isolates from ill calves, and 2 had the same 10-antimicrobial resistance pattern as the other 2 isolates from ill calves.

Herd B had a history of salmonellosis dating back to 2002. Salmonella culture results for all cattle and environmental samples collected during another study on April 12, 2005, and May 9, 2005, were negative. Salmonella Typhimurium was detected in a fecal sample from an apparently healthy calf and an environmental sample from the sick pen collected on June 7, 2005, the date of the first laboratory-confirmed index case of Salmonella shedding. At that time, fecal samples from cattle suspected of having salmonellosis (6 calves and 4 cows) were submitted for bacteriologic culture. Three of the calf samples and all 4 cow samples were positive for Salmonella Typhimurium. All isolates recovered from 4 calf fecal specimens were susceptible to all antimicrobials.

Prevalence of Salmonella shedding—In herd A, bacteriologic culture results of all calves from which samples were obtained between the first laboratory-confirmed index case of Salmonella shedding (January 12, 2005) and the end of the study (March 29, 2005) were included in calculations of the prevalence of fecal shedding of Salmonella spp. Also included were results from 4 calves with clinical signs of salmonellosis and from which samples were collected between January 12, 2005, and January 30, 2005, but that died before the study began (February 7, 2005). Of the 24 calves, 14 (58%) had positive fecal test results for Salmonella Typhimurium variant Copenhagen at least once during the study period. The prevalence of calves shedding this Salmonella serovar at least once within 1 and 2 months after the laboratory-confirmed index case of Salmonella shedding was 75% (12/16) and 58% (14/24), respectively. The prevalence of fecal shedding gradually declined after the first day of the study (26 days after the index case; Figure 1).

View Full Size

Numbers of dairy calves from which isolates of Salmonella spp were (black portion of bars) or were not (white portion of bars) recovered from fecal samples, beginning with the day of detection of the laboratory-confirmed index case of Salmonella shedding (January 12, 2005) in herd A during an outbreak of salmonellosis. Numbers at the top of the bars indicate the percentage of calves that tested positive for Salmonella spp. Beginning 26 days after the index case was identified, samples were collected twice per week.

Citation: Journal of the American Veterinary Medical Association 235, 6; 10.2460/javma.235.6.739

• Download Figure
• Download figure as PowerPoint slide

In herd B, bacteriologic culture results of all calves from which fecal samples were obtained between the first laboratory-confirmed index case of Salmonella shedding (June 7, 2005) and the end of the study (September 4, 2005) were included in calculations of prevalence of fecal shedding of Salmonella spp. Also included was the result of a calf with clinical signs of salmonellosis and from which a fecal sample was collected on June 7, 2005; this calf died before the study began on June 20, 2005. In addition, results were included for fecal samples from 2 healthy calves that were collected on June 7, 2005, although these calves were too old for enrollment in the rest of the study. Of these 82 calves, 25 (30%) had positive fecal test results for Salmonella Typhimurium at least once during the study period. The prevalence of calves shedding Salmonella Typhimurium at least once within 1 and 2 months after the index case was 29% (13/45) and 34% (23/67), respectively. The prevalence of fecal shedding gradually declined after the fifth week of sample collection (Figure 2), and none of the calves were shedding Salmonella spp on November 12, 2005, 158 days after the laboratory-confirmed index case. However, 1 preweaned calf tested on the same date was positive for Salmonella spp, indicating that Salmonella infection was still present in the herd after the study terminated.

View Full Size

Numbers of dairy calves from which isolates of Salmonella spp were recovered from fecal samples beginning with the day of detection of the laboratory-confirmed index case of Salmonella shedding (June 7, 2005) in herd B during an outbreak of salmonellosis. Beginning 13 days after the index case was identified, samples were collected once a week. See Figure 1 for remainder of key.

Citation: Journal of the American Veterinary Medical Association 235, 6; 10.2460/javma.235.6.739

• Download Figure
• Download figure as PowerPoint slide

Antimicrobial resistance patterns—In herd A, 4 antimicrobial resistance patterns were detected in Salmonella isolates recovered from fecal samples collected between January 12, 2005, and March 29, 2005. All isolates were resistant to 9 antimicrobials: amoxicillinclavulanic acid, ampicillin, cefoxitin, ceftiofur, chloramphenicol, kanamycin, streptomycin, sulfisoxazole, and tetracycline. Ten Salmonella isolates recovered from the fecal samples of 8 calves had this resistance pattern. Seventeen isolates from the feces of 10 calves were resistant to those antimicrobials plus ceftriaxone. Three isolates from 1 calf had resistance to the same 9 antimicrobials plus ceftriaxone and TMS. Another isolate from the same calf was resistant to amoxicillin-clavulanic acid, ampicillin, cefoxitin, ceftiofur, chloramphenicol, gentamicin, kanamycin, streptomycin, sulfisoxazole, tetracycline, and TMS. Differences in antimicrobial resistance patterns in Salmonella isolates for individual calves during the study period were detected for 6 calves. For most calves, the difference in resistance patterns between sample collection dates was the acquisition or loss of resistance to ceftriaxone. When temporal changes in the MIC for ceftriaxone were evaluated after controlling for repeated measurements among calves, more isolates from herd A were susceptible to ceftriaxone near the beginning of the study than were isolates recovered later in the study; however, this difference was not significant (fixed-effect partial likelihood test; P = 0.08). One calf accounted for all isolates that were resistant to TMS.

In herd B, results from antimicrobial susceptibility tests indicated that all but 3 of 39 isolates recovered during the study period were susceptible to all antimicrobials. Of the 3 isolates that were resistant to ≥ 1 antimicrobial, 1 was resistant to ampicillin, 1 was resistant to kanamycin and TMS, and 1 was resistant to amoxicillin-clavulanic acid, ampicillin, cefoxitin, chloramphenicol, kanamycin, streptomycin, tetracycline, and TMS. These isolates were detected in fecal samples from 3 calves, and the following week, the isolates recovered from 2 of these calves were susceptible to all antimicrobials. All 3 initially resistant isolates were retested for antimicrobial susceptibility, and results indicated they were susceptible to all antimicrobials.

PFGE patterns—All isolates of Salmonella Typhimurium variant Copenhagen from herd A had the same PFGE pattern. In herd B, all but 1 Salmonella Typhimurium isolate had the same PFGE pattern. The isolate with the unique PFGE pattern was isolated from a calf 55 days after the index case of Salmonella shedding was detected, was not among the 3 isolates in which resistance was detected after the first antimicrobial susceptibility test, and only differed from the predominant PFGE pattern by 1 band. To ensure this band was reproducible, PFGE was performed again, yielding the same results.

Duration of shedding—In herd A, Salmonella spp was isolated from at least 1 fecal sample of 14 calves between January, 12, 2005, and March 29, 2005. The index case of Salmonella shedding was diagnosed when the calf was 18 days old. Most calves (71%) from which Salmonella spp was recovered began shedding Salmonella within 1 to 2 weeks of age. Salmonella spp was isolated more than once from 8 calves, and 4 calves that were fecal positive at least once died before 3 consecutive negative samples could be obtained after the last positive test result. At least 1 negative fecal sample was obtained between the first and last positive fecal sample for 3 calves. The actual duration of Salmonella fecal shedding among all calves ranged from 1 to 29 days, with a median duration of 11 days (Figure 3). The estimated duration of shedding including the unmonitored intervals immediately before and after the actual shedding period ranged from 5 to 35 days, with a median duration of 14 days.

View Full Size

Distribution of duration of Salmonella fecal shedding as detected among dairy calves in 2 herds from which fecal samples were collected twice per week (herd A; black bars; n = 14) or once per week (herd B; white bars; 23) during an outbreak of salmonellosis.

Citation: Journal of the American Veterinary Medical Association 235, 6; 10.2460/javma.235.6.739

• Download Figure
• Download figure as PowerPoint slide

In herd B, Salmonella spp was isolated from at least 1 fecal sample from 23 calves and was recovered from the feces of 10 calves on > 1 sample collection date. Of 12 calves from which fecal samples were obtained within 10 days after birth and for which a positive culture result was obtained at least once during the study, 8 had positive test results within 1 to 2 weeks of age. Culture results were negative for at least 1 sample collected between the first and last Salmonella-positive sample from 5 calves. The actual duration of Salmonella fecal shedding among all calves ranged from 1 to 34 days, with a median duration of 1 day (Figure 3). The estimated duration of shedding, including the unmonitored intervals immediately before and after the actual shedding period, ranged from 7 to 45 days, with a median duration of 9 days.

Serologic response—In herd A, 30% (6/20) of calves from which weekly blood samples were collected and analyzed for serum concentrations of IgG against Salmonella LPS were seropositive (OD > 25%) on at least 1 sampling date. The IgG response did not appear to be attributable to transfer of maternal antibodies because no calves had an OD > 25% until 28 days of age (mean, 40 days; range, 28 to 55 days; Figure 4). Four of the 6 calves had a positive fecal test result before a positive serologic test result. The period between the day at which a positive fecal specimen was collected and a positive serologic result was obtained ranged from 23 to 36 days (median, 29.5 days). One calf had a positive fecal specimen 3 days after the first positive serologic result. Another had Salmonella spp in its feces at 10 days of age but was never seropositive and did not have a measurable serum IgG value until 45 days of age, when the OD was 6%. Because blood samples from calves were collected for only 4 weeks, the serologic response beyond that point could not be assessed.

View Full Size

Changes in serologic response against Salmonella LPS (ELISA %OD) in 6 calves (different symbols) from herd A during an outbreak of salmonellosis. Only calves with at least 1 positive serologic test result (%OD > 25) are indicated. Fecal shedding of Salmonella spp was detected in 5 calves and was not detected in 1 calf (black triangles). Ages (in days) of calves with positive fecal results were as follows: 9 and 17 (diamonds); 16 (squares); 28 and 35 (circles); 10, 14, 21, and 25 (crosses); and 2, 6, 10, and 13 (white triangles).

Citation: Journal of the American Veterinary Medical Association 235, 6; 10.2460/javma.235.6.739

• Download Figure
• Download figure as PowerPoint slide

Seroconversion against Salmonella LPS was not associated with fecal shedding of Salmonella Typhimurium variant Copenhagen among all 20 calves in herd A (Fisher exact test; P = 0.16). Of the 11 calves that shed Salmonella spp at least once during the study, only 5 had a positive serologic response to Salmonella O antigen. The kappa statistic value of 0.32 (95% confidence interval, −0.03 to 0.68) also indicated poor agreement beyond chance between the 2 tests. When the association between the maximum OD and fecal shedding was evaluated, the maximum OD in calves with Salmonella-positive samples (median, 6%; range, 0% to 146%) was not significantly higher (Savage score test; P = 0.11) than the maximum OD in calves with Salmonella-negative samples (median, 0%; range, 0% to 45%).

In herd B, 22% (17/79) of calves from which weekly blood samples were collected and analyzed for serum concentrations of IgG against Salmonella LPS were seropositive (OD > 25%) on at least 1 sample collection date. Seventy-nine percent (13/17) of these calves had high serum IgG concentrations within 1 week after birth that gradually decreased during the study period; this pattern presumably reflected a decrease in passively transferred maternal antibodies (Figure 5). Four calves did not have a high serum IgG concentration until at least 58 days of age (mean, 70.3 days; range, 58 to 83 days; Figure 5). Salmonella spp was isolated from the feces of 3 of these 4 calves. Of the 67 calves that were seronegative (OD > 25%) at their first sample, Salmonella spp was isolated at least once from 80% of calves that became seropositive over the course of the study, compared with 29% of calves that remained seronegative during the study period (Fisher exact test; P = 0.04).

View Full Size

Changes in serologic response against Salmonella LPS (OD as measured with an ELISA) in 15 calves (different symbols) from herd B during an outbreak of salmonellosis. Only calves with multiple serum samples and at least 1 positive serologic test result (OD > 25%) are indicated. Fecal shedding was detected in 5 calves (solid lines) and was not detected in 10 calves (dashed lines).

Citation: Journal of the American Veterinary Medical Association 235, 6; 10.2460/javma.235.6.739

• Download Figure
• Download figure as PowerPoint slide

Blood and fecal samples were collected for at least 4 weeks from most calves in herd B; however, this was not the situation for 11 calves. One calf died after 3 weeks of sample collection, and 10 calves enrolled within the last month of the study contributed samples for only 1 to 3 weeks before the study ended. Therefore, for the analysis of the association between fecal shedding of Salmonella spp and a subsequent seropositive test result, data from these 11 calves were excluded. Those included in that analysis were the 56 calves that were seronegative when the first blood sample was collected and from which samples were collected at least 4 times. Results indicated that only 19% (4/21) of calves that shed Salmonella spp during the study seroconverted; this percentage was not significantly higher among calves with positive fecal test results, compared with those with negative results (3% [1/35]) (Fisher exact test; P = 0.06). However, the kappa statistic indicated poor agreement beyond chance between the 2 tests (N = 0.19; 95% confidence interval, 0.01 to 0.40).

Among 10 calves tested at least 4 times, 1 was seropositive within the first week after birth and shedding Salmonella spp when subsequent samples were collected, compared with 25% (8/32) of calves that were seronegative; however, this difference was not significant (Fisher exact test; P′ = 0.40). When percentage OD was treated as a continuous variable instead of a dichotomous one (seronegative or seropositive), no association was detected between a high serum concentration of IgG against Salmonella LPS within the first week after birth and subsequent fecal shedding of Salmonella spp; the median OD for blood samples obtained within the first week after birth for both fecal positive and fecal negative calves was 14% (Savage score test; P = 0.5).

Discussion

Salmonella Typhimurium (including variant Copenhagen) was the most common serotype isolated from specimens from clinically infected humans and the second most common serotype isolated from specimens from clinically infected cattle in 2004.²⁸ Because Salmonella spp can be recovered from dairy cattle for up to 3.5 years after clinical infection,^6,7,29 the importance of cattle as a reservoir for Salmonella spp resulting in the subsequent transmission to humans is a subject of continuous debate, particularly with respect to the emergence and dissemination of antimicrobial-resistant strains. Even so, the dynamics of Salmonella infection in dairy cattle herds is poorly understood. To improve this understanding, we monitored fecal Salmonella shedding in 2 dairy herds that underwent outbreaks of clinical salmonellosis. The origin of the outbreak strains was unknown. To rule out the possibility that these Salmonella strains were introduced into the herds from other herds while the herds were involved in another study, the Salmonella serotypes from other herds visited at the same times as the present herds for sample collection were determined, indicating that Salmonella Typhimurium was not isolated from those other herds during that period.

Mathematic infectious disease models have been constructed to explain the dynamics of Salmonella infections in dairy herds, with the ultimate goal of developing more effective control strategies.^30,31 However, these models must make use of data obtained through experimental trials because data concerning natural infections are sparse, and this experimental data may poorly represent the dynamics of natural infection. In the present study, data from 2 natural outbreaks of salmonellosis were prospectively collected. Results indicated that most calves in both herds were only positive for fecal shedding of Salmonella spp at 1 sample collection point. Because fecal samples were not collected every day, the exact start and end time of fecal shedding was unknown. After statistically accounting for an unmonitored shedding period immediately before and after the actual shedding period, the median duration of shedding increased by 3 days in farm A and by 8 days in farm B. The sample collection schedule used to estimate the duration of shedding may have overestimated the true duration of shedding because calves that shed Salmonella spp for a short period, before enrollment in the study or between sample collection dates, would not be detected. In addition, calves that shed Salmonella organisms in low concentrations may not have been detected because sensitivity of bacteriologic culture of feces is dependent on the fecal concentration of the organism. In our laboratory, the sensitivity of bacteriologic culture of feces for detection of Salmonella spp is approximately 100% and 50% for concentrations of at least 10⁴ and 10¹ colony-forming units/g of feces, respectively (unpublished data). The use of rectal swabs for sample collection also may have decreased the diagnostic sensitivity of bacteriologic culture in our study.

One study¹¹ of fecal shedding of Salmonella Typhimurium by experimentally infected calves revealed that duration of shedding ranged from 3 to 20 days, depending on the inoculating dose and breed of calf. The difference in the duration of shedding between the results of the present observational study and those of the experimental study could be attributable to several factors, including differences in frequency of sample collection, duration of follow-up, number of organisms ingested by each calf, and virulence of the infecting Salmonella strain. It should also be recognized that bacteriologic culture of fecal samples did not differentiate between calves colonized with Salmonella spp and those that ingested and passed the organism in the feces without colonization.

Because the present study was conducted in commercial herds, treatment and management decisions made by herd veterinarians and owners may have affected the course of the outbreaks. After the initial clinical cases of salmonellosis were identified, a commercial vaccine against Salmonella Newport was administered to cattle in both herds. However, so far no effects of this vaccine on fecal shedding of Salmonella or clinical salmonellosis have been detected.³² All calves from both herds were also treated with TMS after the outbreaks began. Resistance to TMS is uncommon among Salmonella spp,³³ but was detected in a fecal isolate from 1 calf in herd A. The emergence of TMS resistance following whole-herd treatment during an outbreak of Salmonella Newport infection in a calf-rearing facility emphasizes the importance of judicious use of antimicrobials,³⁴ particularly because the efficacy of this extralabel treatment has not been established. In the present study, the effect of antimicrobial treatment and other interventions on the course of the outbreaks could not be determined.

The prevalence of antimicrobial-resistant Salmonella organisms on dairy farms could be related to local selection pressures (antimicrobial use) or the emergence of resistant clonal strains that are successfully disseminated within and between populations (eg, the emergence and worldwide spread of multidrug-resistant Salmonella Typhimurium DT104 in the 1990s).^35,36 The antimicrobialresistance profiles of isolates from herd A suggested that a combination of these factors may have affected the prevalence of antimicrobial-resistant Salmonella isolates in that herd. Herd A did not have a recent history of salmonellosis before the initial laboratory-confirmed index case of Salmonella shedding was detected. In addition, Salmonella spp was not recovered from any of the fecal samples obtained from the environment and healthy cows and calves in the 3 months prior to the outbreak.

Two purchased cows were introduced to herd A in the year preceding the outbreak, but these cows were not tested for Salmonella spp, so it is not known whether Salmonella spp was introduced through the purchase of infected cattle. The purchase of cattle was associated with Salmonella infection in another study.³⁷ All Salmonella isolates from herd A, even at the beginning of the outbreak, were resistant to multiple antimicrobials, which supports the supposition that a multidrug-resistant strain had been introduced, although its origin was unknown. As the outbreak progressed, the prevalence of isolates with resistance to ceftriaxone increased. Antimicrobial use may have contributed to this increase because ceftiofur, which is also a cephalosporin with broad-spectrum activity, was used in the treatment of neonatal diarrhea and respiratory disease.

All Salmonella isolates from herd A shared the same PFGE pattern. Pulsed-field gel electrophoresis has been used to characterize Salmonella subtypes from human and animal sources and in outbreak investigations, and differences in the discriminatory power of PFGE have been reported for various serovars.^38–40 In the present study, PFGE was used because it is typically highly discriminatory and because a standardized protocol²² is available that allows for generation of subtyping data, which can be compared between studies. In an investigation of an outbreak of salmonellosis in humans, PFGE was used to confirm that Salmonella Typhimurium isolates with different antimicrobial resistance patterns are genetically distinct.⁴¹ In the present study, the finding that all isolates from herd A shared the same PFGE pattern suggested that it was unlikely that the introduction of a unique ceftriaxone-resistant Salmonella strain accounted for the increase in ceftriaxone resistance. However, limited discrimination by PFGE for certain Salmonella serovars could also account for detection of only 1 PFGE pattern in herd A. All except 1 isolate shared the same PFGE pattern in herd B. The extra band present in that unique isolate most likely indicated that it represented a progeny of the predominant strain, possibly through acquisition of a plasmid, a prophage change, or another type of mutation.

The importance of humoral immunity in protection against Salmonella infection in calves is not well understood. In the present study, most calves that seroconverted had Salmonella spp detected in their feces (5 of 6 calves in herd A and 4 of 5 calves in herd B), but overall there was not a strong association between fecal shedding and a serologic response. For example, among calves initially seronegative, only 45% and 19% of those with positive fecal test results seroconverted in herd A and herd B, respectively. Other studies have revealed an increase in the antibody response among calves after exposure to Salmonella Typhimurium. In 1 such study,¹⁶ the cellular and humoral immune responses of 2- to 4-month-old calves that shed Salmonella Typhimurium within the previous 6 to 10 weeks were compared with responses of calves from uninfected herds. Investigators in that study reported significantly higher serum titers of antibody against Salmonella LPS in infected versus uninfected calves. In addition, 2 experimental studies^42–44 in which calves from Salmonella-free herds were vaccinated with a modified-live aromatic-dependent Salmonella Dublin bacterin or a live Salmonella Typhimurium auxotrophic mutant strain revealed an increase in serum titers of antibody against Salmonella LPS in both studies. Calves from those studies were between 1 and 5 weeks of age at the time of vaccination and were monitored for 4 to 6 weeks. Differences between the results of our study and those of other studies on the serologic response against and fecal shedding of Salmonella spp^8,43 could be related to the effects of age, duration of follow-up, and the number of salmonellae consumed by calves. Fecal shedding without substantial colonization could also account for a lack of association with seroconversion. It is also possible that the small sample size in herd A could have reduced the power to detect an association.

In the present study, it was not likely that the vaccine^a used resulted in the changes in serum antibody concentrations detected because that vaccine is based on different antigens than those in the ELISA. Whereas some calves from herd B had a high OD within the first week after birth, no calves in dairy A had a high OD at that point. This was most likely attributable to passive transfer of maternal antibodies and suggested that the dams of some calves in dairy B may have had recent exposure to Salmonella Typhimurium or cross-reacting serotypes, whereas the dams of the calves in herd A probably did not. Although subsequent Salmonella fecal shedding appeared to be less common among calves with high serum IgG values within 1 week after birth, the effect was not significant.

The effects of colostral immunity (by feeding calves colostrum from cows vaccinated with an inactivated Salmonella vaccine) on fecal shedding of Salmonella spp following oral challenge in calves were examined in an experimental study.¹⁵ Although all calves in that study shed Salmonella spp immediately after dosing, there was a difference in somatic antibody titers between calves fed colostrum from vaccinated cows and calves fed colostrum from unvaccinated cows. In addition, the number of salmonellae shed decreased more quickly in calves fed colostrum from vaccinated cows, compared with the rate in calves fed colostrum from unvaccinated cows.¹⁵ The investigators in that study suggested that the protective effects of colostral antibodies were attributable to a local gastrointestinal effect. In our study, there was only 1 calf with an OD > 25% at birth that subsequently shed Salmonella spp at least once during the study, thereby limiting the power to detect an association between seropositivity at birth and the duration of Salmonella fecal shedding.

In the study reported here, although the 2 herds differed in serologic response to an outbreak of salmonellosis and the outbreak strains were distinct by serotype, antimicrobial resistance profile, and PFGE pattern, the duration of fecal shedding of Salmonella spp was similar between herds. Most calves shed Salmonella spp only once, and few calves shed for > 14 days. The multidrug-resistant Salmonella strain identified in herd A appeared to become less susceptible to ceftriaxone as the outbreak progressed.