Farm-Level Associations with the Shedding of Salmonella and Antimicrobial-Resistant Salmonella in U.S. Dairy Cattle

Abstract

Salmonella enterica is the leading cause of foodborne-related deaths and hospitalizations within the United States. Infections caused by antimicrobial-resistant (AMR) strains are associated with higher hospital costs and case fatality. The objective for this study was to determine the association of management practices with the recovery of Salmonella and AMR Salmonella on dairy herds. Individual adult cow fecal samples and/or composite fecal samples were collected from 265 dairy herds in 17 states. Samples were cultured for Salmonella, and the MIC was determined for 15 antimicrobials. Herds were classified as Salmonella positive if at least one isolate was recovered, and AMR Salmonella positive if at least one resistant isolate was recovered. Questionnaires regarding management practices were administered to herd operators, and a subset of practices was selected based on subject knowledge and prior research. Data on preventive and therapeutic antimicrobial usage were included in the analysis. Logistic regression models were used to determine which practices were significantly (p<0.05) associated with each herd classification. A total of 124 and 25 herds were classified as Salmonella positive and AMR Salmonella positive, respectively. Variables significantly associated with Salmonella-positive herds included using sprinklers or misters for heat abatement (OR=2.8; CI: 1.6–4.9), feeding anionic salts to cows (OR=1.9; CI: 1.1–3.5), and feeding ionophores to cows (OR=2.1; CI: 1.2–3.7). Herds that used a broadcast/solid spread had lower odds (OR=0.26; CI: 0.11–0.63) of being Salmonella positive. Herds with at least one resistant isolate were more likely to have used composted/dried manure for bedding relative to herds with only susceptible isolates (OR=3.6; CI: 1.2–11.0). These results can be useful to focus additional research aimed at decreasing the prevalence of Salmonella and AMR Salmonella on U.S. dairy herds.

Introduction

S almonella enterica is the most frequent cause of foodborne-related deaths and hospitalizations in humans within the United States (Scallan, 2011), and an important cause of clinical disease in cows and calves on dairy farms (Cummings et al., 2009). Antimicrobial resistance (AMR) in Salmonella inhibits the ability of physicians and veterinarians to treat severe infections, and results in increases in health care costs and mortality in human patients (Maragakis et al., 2008). Salmonella are symptomatically or asymptomatically excreted by dairy cattle and are present in the environment of many dairy farms (Cummings et al., 2009; Ruzante et al., 2010). The serovars, resistance phenotypes, and genetic subtypes of Salmonella isolated from dairy cattle have considerable overlap with those that cause disease in humans (Alcaine et al., 2006; Wedel et al., 2005; Zhao et al., 2003). Reductions in the prevalence and AMR of Salmonella on dairy farms may lead to subsequent reductions in the incidence and/or severity of salmonellosis in humans.

Prior cross-sectional studies have shown that the presence or absence of Salmonella has been consistently associated with farm-level characteristics, including herd size (Huston et al., 2002; Kabagambe et al., 2000; Ruzante et al., 2010; Wells et al., 2001). Likewise, the prevalence of AMR on dairy farms was unevenly distributed among regions and farm types (Berge et al., 2010). These uneven distributions suggest farm-level management practices may be associated with the presence of Salmonella and AMR Salmonella. Indeed, dairy farm management practices can alter the environmental niche to create a more or less favorable environment for Salmonella. Associations with farm Salmonella status in prior research have included the use of flush water systems and brewer's products in rations (Kabagambe et al., 2000), lack of tie stall usage, lack of enclosed building for feed storage, cow access to surface water, disposal of manure in a liquid form, and lack of ionophore inclusion in weaned calf or bred heifer diets (Fossler et al., 2005a).

Research on management practices associated with AMR in bacteria on dairy farms has primarily focused on antimicrobial usage (Kaneene et al., 2009; Ray et al., 2006). However, there may be non-antimicrobial management practices that can be altered to impact the ability of AMR bacteria to compete with their susceptible counterparts (Khachatryan et al., 2006). Given the lack of consensus on the net effect of mandated reductions in agricultural antimicrobial usage, alternative interventions for reducing the prevalence of AMR Salmonella should continue to be explored.

The objective of this cross-sectional study was to compare the use of management practices between Salmonella-positive and Salmonella-negative farms, and between AMR Salmonella-positive farms relative to farms where only pansusceptible Salmonella were isolated. The hypothesis tested by this research is that modifiable management practices are significantly associated with the recovery of Salmonella and AMR Salmonella on U.S. dairy farms.

Methods

Data collection

The 2007 National Animal Health Monitoring System (NAHMS) Dairy Study was conducted as a cooperative effort between the Animal and Plant Health Inspection Service (APHIS) and the National Agricultural Statistics Service (NASS). Detailed descriptions of methods are available elsewhere (USDA, 2007). Briefly, 17 states were chosen to represent 79.5% of the U.S. dairy operations and 82.5% of the dairy cow population. States included those from the West (Idaho, New Mexico, Texas, and Washington) and from the East (Indiana, Iowa, Kentucky, Michigan, Minnesota, Missouri, New York, Pennsylvania, Vermont, and Wisconsin). At the outset, a stratified random sample with unequal selection probabilities was used to select farms based on herd sizes of 1–99, 100–499, or >500 milking cows.

In Phase I of Dairy 2007, NASS enumerators administered questionnaires to 2,194 operations between January 1 and January 31, 2007. Of 2,194 operations that completed Phase I, 1,077 consented to be contacted by veterinary medical officers for more information on the study. Of the 1,077 operations, 582 agreed to continue with Phase II of the study, and state and federal Veterinary Medical Officers (VMOs) and Animal Health Technicians (AHTs) administered two questionnaires between February 26 and August 31, 2007. Categories included cattle inventory, general management, health management, housing, biosecurity, antibiotic use, and nutrient management.

A subset of eligible farms that participated in the both phases of the study was selected for composite fecal sampling. A subset that underwent composite fecal sampling was selected for individual collection. Farms were visited once by state and federal VMOs and AHTs between February 28, 2007 and August 30, 2007. Up to 35 individual samples (regardless of herd size) were collected via rectal retrieval, using individual sleeves. This sample size provides 95% confidence of detecting a positive animal assuming a prevalence not less than 8%. Up to five sick cows and five cows scheduled for culling were sampled, and the remainder were from cows with saleable milk. Composite samples were collected from adult cow areas where manure accumulated. Recommended sites included alleyways, pens, exits from parlors, floors of holding pens, flush water, gutter cleaner, lagoons or manure pits, and manure spreaders. The six composite fecal samples were collected from six different adult cow areas. Each composite fecal sample consisted of approximately 4 oz of manure/slurry from six different locations within a cow area. Samples were placed in sterile bags and placed on ice for shipment to the ARS Bacterial Epidemiology and Antimicrobial Resistance Research Unit (BEAR, Athens, GA).

Laboratory methods

Up to four typical Salmonella colonies were isolated from samples using previously described methods (USDA, 2010). Serovar identification was performed using standard agglutination techniques. Where all isolates within the same sample had the same serogroup, only one isolate underwent AMR testing.

The minimum inhibitory concentration (MIC) of 15 antimicrobials was determined using the TREK Sensititre^® automated testing system (TREK Diagnostic Systems, Cleveland, OH). The tested antimicrobials included amikacin, amoxicillin-clavulanic acid, ampicillin, cefoxitin ceftiofur, ceftriaxone, chloramphenicol, ciprofloxacin, gentamicin, kanamycin, nalidixic acid, streptomycin, tetracycline, and trimethoprim-sulfamethoxazole. Isolates were classified as resistant, intermediate, or susceptible according to guidelines established by the Clinical and Laboratory Standards Institute (CLSI, 2008). Where unavailable, breakpoints used by the National Antimicrobial Resistance Monitoring System (NARMS) were used (FDA, 2011).

Statistical analysis

Variables plausibly associated with Salmonella shedding or AMR in Salmonella were chosen from the questionnaires based on subject matter knowledge and prior research. Data on preventive antimicrobial use included medicated milk replacer, in-feed antimicrobial use, and dry-cow intramammary therapy. Therapeutic antimicrobial usage questions were collapsed into binary variables that reflected use of each class (cephalosporin, beta-lactam, macrolide, aminoglycoside, sulfonamide, florfenicol, and tetracycline) within the prior 12 months. Separate sets of variables were created within each animal type (preweaned heifers, weaned heifers, or adult cows).

Farms were considered Salmonella positive if at least one of the individual fecal or composite samples was positive, and farms were considered AMR Salmonella positive if at least one isolate had an MIC that exceeded the respective breakpoint. SAS^® software (version 9.1; SAS Institute, Cary, NC) was used to construct separate logistic regression models for each outcome, using identical model building approaches. The same subset of variables was used for both models. The objective for the second model was to compare farms where AMR Salmonella were recovered relative to farms where only susceptible Salmonella were found. Therefore, only Salmonella-positive farms were included in the second model.

Effect estimates and standard errors were determined using maximum likelihood estimation, and the significance was assessed using the Wald test. Variables with a univariable association where p was <0.15 were included as candidates in a forward stepwise variable selection procedure, and were retained in the final model if p was <0.05. Collinearity and plausible interactions were assessed using the Pearson correlation coefficient and Breslow-day tests, respectively. Potential confounding by herd size and season was assessed post-hoc by comparisons of the crude and adjusted ORs.

Results

Farm participation and classification

Samples were collected from 265 dairy farms in 17 states. Composite samples were collected from 260 farms, and individual samples were collected from 121 farms. Of 144 farms where only composite samples were collected, five or six samples were collected on 136 farms, and of the farms where both sample types were collected, a total of 31–48 samples were collected. At least one Salmonella isolate was recovered on 124 farms (46.8%) (Table 1). For the majority (78%) of positive farms, at least one isolate was recovered from greater than 10% of the samples. Among Salmonella-positive farms, 25 (20.2%) farms had at least one resistant isolate (9.4% of all farms), and were classified as AMR positive. Multi-drug resistant (MDR) (resistance to >2 antimicrobials) and penta-resistant isolates (any five antimicrobials) were recovered from 22 and 15 farms, respectively. All isolates classified as MDR were resistant to >1 class of antimicrobials.

Table 1.

Association of Herd Size and Region with the Presence of Salmonella ^a or the Presence of AMR (+) Salmonella ^b Among U.S. Dairy Operations Sampled in 2007

	Category	Number of herds	Number (%) of all herds that were Salmonella positive	P-value	Number (%) of Salmonella-positive herds that were AMR (+)	p-value
All		265	124 (46.8)		25 (20.2)
Herd size	<100	93	33 (35.5)	0.004	3 (9.1)	0.177
	100–500	95	44 (46.3)		11 (25.0)
	>500	77	47 (61.0)		11 (23.4)
Region	West	53	18 (34.0)	0.036	6 (33.3)	0.132
	East	212	106 (50.0)		19 (17.9)

At least one Salmonella-positive fecal sample recovered from individual cow or environmental samples.

At least one Salmonella isolate was resistant to any of the antimicrobials tested, among herds where Salmonella was recovered.

Risk factors for the presence of Salmonella on dairy operations

A total of 142 variables were selected from the available data. The odds of being Salmonella positive were significantly higher for larger farms and farms in the Eastern region of the United States (Table 1). A larger proportion of farms sampled in June, July, or August were positive relative to farms sampled in the late winter and spring (February through May). Forty-one variables where p was <0.15 were considered for selection in the final model (Table 2). The final model indicated that farms that reported the use of sprinklers or misters for heat abatement (OR=2.8; CI: 1.6–4.9), feeding anionic salts to cows close to calving (OR=1.9; CI: 1.1–3.5), and feeding ionophores to lactating cows (OR=2.1; CI: 1.2–3.7) had higher odds of being Salmonella positive. Farms that reported manure application by broadcast/solid spreader had lower odds (OR=0.26; CI: 0.11–0.63) of being Salmonella positive (Table 3). There were no significant interactions between herd size, region, and variables in the final model. Inclusion of herd size resulted in <15% difference between adjusted and unadjusted ORs (Table 4). Likewise, season (spring or summer) was significant in the final model (p<0.05), but had a small effect (<7% difference) on the ORs for the management practices in the final model. The Pearson correlation coefficients for the variables in the final model were all <0.15.

Table 2.

Univariable Association of Selected Management Practices ^a with the Presence of Salmonella on U.S. Dairy Farms Sampled in 2007

	Number (%) of all herds positive for Salmonella by questionnaire response
Management practice	Yes		No		p-value
Disposable or clean boots for visitors	72	(61)	52	(35.4)	<0.001
Sprinklers or misters for heat abatement in lactating cows	64	(61.5)	60	(37.3)	<0.001
Weaned dairy heifers raised on the operation	100	(42.9)	24	(75)	0.001
Subsurface injection of manure	31	(67.4)	87	(41.8)	0.002
Whole cottonseed fed to lactating cows	81	(54.7)	42	(36.2)	0.003
Bred dairy heifers raised on the operation	102	(43.4)	22	(73.3)	0.003
Anionic salts fed to close-up cows	52	(59.1)	71	(40.3)	0.004
Lab-confirmed clinical Salmonella case	27	(67.5)	94	(42.5)	0.005
Dairy replacements purchased	21	(72.4)	103	(43.6)	0.005
Manure application by broadcast/solid spreader	95	(43)	23	(69.7)	0.006
Any cattle brought onto the operation	71	(55.5)	53	(38.7)	0.007
Offsite heifer grower used	37	(61.7)	87	(42.4)	0.010
Ionophores fed to cows	69	(56.1)	53	(38.4)	0.017
Vaccines for E. coli mastitis in heifers	57	(55.9)	67	(41.1)	0.020
Johne's cows usually enter calving pen	22	(66.7)	51	(48.6)	0.029
Siderophore receptors and porins (SRP) vaccines	21	(70)	2	(50)	0.030
Fans used for heat abatement in lactating cows	102	(50.5)	22	(34.9)	0.032
Manure applied to actively growing grain or oilseed crops	24	(61.5)	94	(43.5)	0.041
Tetracylines for treatment in cows	61	(54)	63	(41.4)	0.044
Brewery byproducts fed to lactating cows	63	(53.4)	60	(41.1)	0.047
Gentamicin for treatment in cows	7	(87.5)	117	(45.5)	0.048
Unpasteurized whole milk fed to calves	13	(31.7)	96	(48.7)	0.050
Surface application of manure by tank wagon or tank truck	65	(52.4)	53	(40.8)	0.063
Bred dairy heifers brought onto the operation	36	(62.1)	29	(46)	0.069
Vaccines for E. coli mastitis in cows	77	(51.7)	47	(40.5)	0.072
Brewery byproducts fed to dry cows	39	(55.7)	84	(43.3)	0.076
Cows housed in tie-stalls	19	(35.8)	105	(49.5)	0.076
Antimicrobials in weaned heifer rations	75	(49)	36	(37.9)	0.088
Sick cows usually enter calving pen	30	(38.5)	92	(50)	0.088
Farm participates in a Johne's program other than a state-sponsored program	13	(65)	111	(45.3)	0.097
Macrolides for treatment in calves	30	(56.6)	94	(44.3)	0.111
Farm participates in a Johne's control program (not state-sponsored) developed specifically for the farm	32	(56.1)	92	(44.2)	0.112
Sulfonamides for treatment in calves	12	(34.3)	112	(48.7)	0.115
Novobiocin dry cow intramammary therapy	1	(14.3)	123	(47.7)	0.118
Dry cow intramammary therapy	120	(48)	4	(26.7)	0.119
Probiotics in dairy cows	52	(53.6)	70	(44)	0.119
Vaccines for Salmonella in cows	35	(58.3)	5	(45.5)	0.126
Pen G dry cow intramammary therapy	3	(25)	121	(47.8)	0.136
Anionic salts fed to springing heifers	33	(55)	90	(44.1)	0.139
Tetracylines for treatment in weaned heifers	29	(55.8)	95	(44.6)	0.150
	All	Some	No incoming cattle	None
Aware of the source of incoming cattle	48 (44.0)	21 (70.0)	47 (43.1)	8 (53.3)	0.067

Management practices included are those with a univariate p-value of <0.15.

Table 3.

Results of the Final Logistic Regression Model for the Association of Management Practices with the Presence of Salmonella in Cow Fecal Samples Among U.S. Dairy Operations Sampled in 2007

Variable	Level	Estimate	SE	p-value
Sprinklers or misters for heat abatement in lactating cows	YesNo (ref)	0.521	0.142	<0.001
Anionic salts fed to close-up cows	YesNo (ref)	0.334	0.147	0.023
Manure application by broadcast/solid spreader	YesNo (ref)	−0.666	0.222	0.003
Ionophores fed to cows	YesNo (ref)	0.375	0.139	0.007

SE, standard error.

Table 4.

Odds Ratios for Management Practices Included in the Final Logistic Regression Model of the Presence of Salmonella on U.S. Dairy Operations Sampled in 2007: Results With and Without Adjustment for Herd Size Category Are Included

Variable	Level	OR (95% CI)		OR (95% CI)
Herd size	<100			0.695	(0.32–1.492)
	100–500>500 (ref)			0.72	(0.362–1.43)
Sprinklers or misters for heat abatement in lactating cows	YesNo (ref)	2.832	(1.62–4.94)	2.527	(1.378–4.633)
Anionic salts fed to close-up cows	YesNo (ref)	1.948	(1.096–3.463)	1.840	(1.023–3.309)
Manure application by broadcast/solid spreader	YesNo (ref)	0.264	(0.111–0.630)	0.263	(0.110–0.631)
Ionophores fed to cows	YesNo (ref)	2.118	(1.228–3.653)	2.070	(1.194–3.588)

Risk factors for the presence of antimicrobial-resistant Salmonella

There was not a significant association between herd size and region with the presence of AMR Salmonella compared to farms with only susceptible Salmonella (p>0.15; Table 1). Seventeen management practices with a univariable association (p<0.15) were considered for the final model (Table 5). The selection procedure retained only a single variable, which was the use of composted manure for bedding, (OR=3.6; CI: 1.2–11.0).

Table 5.

Univariable Association of Selected Herd Management Practices ^a with the Presence of at Least One Antimicrobial-Resistant (AMR) Salmonella Among Salmonella-Positive Dairy Farms in the United States in 2007

	Number (%) of AMR (+) herds by questionnaire response
Management practice	Yes		No		p-value
Manure applied to actively growing pasture or hay	6	(10.3)	17	(28.3)	0.017
Composted/dried manure for bedding in lactating cows	7	(43.8)	18	(16.7)	0.017
Whole cottonseed fed to lactating cows	21	(25.9)	4	(9.5)	0.04
Manure application by broadcast/solid spreader	15	(15.8)	8	(34.8)	0.045
Lactating cows brought onto the operation	2	(7.4)	11	(28.2)	0.051
Cottonseed meal or hulls fed to lactating cows	6	(40)	19	(17.6)	0.051
Sprinklers or misters used	17	(26.6)	8	(13.3)	0.071
Cephalosporins for treatment in calves	9	(32.1)	16	(16.7)	0.078
Surface application of manure by tank wagon or tank truck	9	(13.8)	14	(26.4)	0.091
Same equipment for handling feed and manure	3	(11.1)	9	(16.4)	0.097
Dairy replacements purchased from a dealer	7	(33.3)	18	(17.5)	0.106
Anionic salts fed to close-up cows	7	(13.5)	18	(25.4)	0.111
Ionophores in dairy replacement heifers	13	(14.4)	8	(27.6)	0.112
Disposable or clean boots for visitors	18	(25)	7	(13.5)	0.119
Macrolides for treatment in calves	3	(10)	22	(23.4)	0.123
Florfenicol for treatment in weaned heifers	17	(17.3)	8	(30.8)	0.135
Shared heavy equipment with other operations	7	(13.7)	18	(24.7)	0.14
	West		East
Region of the United States	6	(33.3)	19	(17.9)	0.139

Management practices listed are those with a univariate p-value of<0.15.

Discussion

The 265 dairy farms in this study population enable broad inferences to dairy farms in the United States and also provide the statistical power necessary to study association of practices implemented at the farm-level with Salmonella and AMR Salmonella.

The consideration of a large numbers of variables increases the probability of chance associations (Dohoo et al., 2009). Also, this cross-sectional study is unable to assess the temporality of the risk factors and outcomes. Furthermore, the categorizations used in this study design may oversimplify the complex epidemiology and ecology of Salmonella and AMR on dairy farms. Misclassification of herds with respect to risk factors is problematic for exposures that require recollection; however, these misclassifications are expected to be non-differential between positive and negative herds and are therefore likely to bias the results towards the null.

Despite these limitations, this study effectively describes management practice differences between herds that were Salmonella and AMR Salmonella positive and negative in the NAHMS Dairy 2007 Study. These results are useful to confirm prior findings and generate hypotheses to focus additional research to identify effective pre-harvest food safety interventions.

Herd size

Farms with >500 cows had higher odds of being Salmonella positive than smaller herds (Table 1). This association has been found in prior studies of Salmonella on dairy farms (Blau et al., 2005; Huston et al., 2002; Warnick et al., 2001), in other livestock populations (Gardner et al., 2007), and for other bacterial species (USDA, 2011), suggesting that there are inherent differences in transmission dynamics between different herd sizes (Gardner et al., 2007). Herd size was concurrently associated with the use of specific management practices and the Salmonella status of the herd, indicating a confounding effect. Adjusting the estimates of the variables in the final model for herd size did not have an important impact on effect estimates or the interpretation of the results (Table 4).

Management practices

Farms that used anionic salts had higher odds of being Salmonella positive. Anionic salts may increase Salmonella shedding through negative effects on feed intake prior to parturition. Alternatively, changes in pH and mineral concentrations in the gut may provide a more or less favorable environment for Salmonella. Although rumen and blood pH are lower in cows on DCAD diets (Goff, 2000), higher absorption of anions in the rumen relative to cations may result in a higher pH in the lower gut. Regardless, the well known positive impacts of anionic salts on the incidence of peri-parturient hypocalcemia and cow health must be weighed against any potential effects on Salmonella shedding.

Farms that used sprinklers and misters for heat abatement during the summer months had higher odds of being Salmonella positive. This association may be confounded by a higher frequency of use of heat abatement practices in warm regions, where Salmonella are more apt to survive and replicate within the environment. Alternatively, wetter farm environments may facilitate survival and/or replication. The use of fans had a significant univariable association with the Salmonella status of herds (p=0.03) (Table 2) but was not significant in the final model. Longitudinal research with more intensive sampling within herds is needed to understand variations in shedding with changing conditions. On-farm trials may also be useful to examine the effect of heat abatement practices on within-farm variations in shedding.

Salmonella shedding has been shown to be higher in the spring, summer, and fall relative to winter (Fossler et al., 2005). In this study, sampling took place from late February through August. A larger proportion of the herds sampled in the summer (June through August) were positive relative to herds sampled from February through May. While this is a potential source of confounding, season (spring or summer) was not significantly associated (p>0.05) with variables in the final model, and season did not have an important effect on the ORs for variables in the final model (<7% difference).

Farms that used a broadcast or solid spreader had lower odds of being Salmonella positive. This is consistent with a prior longitudinal study (Fossler et al., 2005a) and with experimental work that has shown longer persistence of Salmonella in manure slurry relative to static manure piles (Nicholson et al., 2005; Toth et al., 2011). Manure in a liquid form may also be dispersed more broadly, facilitating ingestion and colonization (Fossler et al., 2005b). The consistency across experimental and observational research strengthens the potential causal association between manure management practices and Salmonella recovery.

Farms that used ionophores had twice the odds of being Salmonella-positive. Prior trials and observational studies with ionophores have not shown a significant association with Salmonella shedding (Edrington et al., 2003a,b, 2006). In fact, the opposite effect on Salmonella shedding has been shown (Fossler et al., 2005b). Given the conflicting research and lack of evidence in experimental work, the association in our research may be a chance or spurious association due to unmeasured confounders.

Antimicrobial-resistant Salmonella

The stepwise selection procedure retained only the single most significant variable from the univariable analysis. The differences in statistical significance among the practices in univariable analysis are small. Nonetheless, the univariable results are useful to demonstrate the relative strength of associations of management practices with AMR Salmonella. Notably, three of the four most significant practices involved manure management, suggesting these practices have important impacts on both the presence and absence of Salmonella and selection resistant populations (Table 5).

There were no significant (p<0.05) univariable associations between antimicrobial use and the AMR farm classifications (Table 5). Cephalosporin use in calves, macrolide use in calves, and florfenicol use in weaned heifers tended (p<0.15) to be more frequent on AMR- positive farms relative to farms with only susceptible Salmonella. These weak associations are consistent with research that has found relatively small differences in the AMR of Salmonella between organic (no antimicrobial use) and conventional farms (Ray et al., 2007). Additionally, decreases in the within-farm estimates of AMR of Salmonella over time have been reported despite concurrent on-farm antimicrobial usage (Habing et al., 2012). These results may reflect a larger influence of advantageous fitness genes relative to AMR genes for competition and survival of Salmonella in dairy cows and the dairy farm environment. However, the low proportion of farms with AMR Salmonella limits the statistical power. Additionally, the AMR classification may be an oversimplified construct for the complex epidemiology of AMR. However, separate analyses for resistance in each antimicrobial would unacceptably curtail the statistical power for farm-level analyses. The lack of quantitative usage data and the lack of recording of changes in use over time could be expected to bias the associations towards the null. Furthermore, the majority of farms in this study used at least one antimicrobial class. Therefore, this study is not suited to compare farms with and without antimicrobial use. Rather, these results are useful to examine the relative strength of association of use of different antimicrobial class in different animal types with the AMR classification.

Conclusion

This study identified management practices that were significantly associated with the presence or absence of Salmonella and AMR Salmonella on dairy farms. Management practices in the final model for Salmonella status varied with respect to their consistency with prior research, but may represent modifiable pre-harvest food safety targets. Manure management practices were significantly associated with Salmonella and AMR Salmonella. These results will be useful to focus additional research aimed at identifying interventions to reduce the reservoir of Salmonella on dairy farms.

Footnotes

Acknowledgments

We would like to acknowledge the technical assistance of Sandra House.

Disclosure Statement

No competing financial interests exist.

References

Alcaine

, Soyer

, Warnick

, Su

, Sukhnanand

, Richards

, Fortes

, McDonough

, Root

, Dumas , Grohn

, Wiedmann

. Multilocus sequence typing supports the hypothesis that cow- and human-associated Salmonella isolates represent distinct and overlapping populations. Appl Environ Microb, 2006; 72:7575–7585.

Berge

, Hancock

, Sischo

, Besser

. Geographic, farm, and animal factors associated with multiple antimicrobial resistance in fecal Escherichia coli isolates from cattle in the western United States. J Am Vet Med Assoc, 2010; 236:1338–1344.

Blau

, McCluskey

, Ladely

, Dargatz

, Ferdorka-Cray

, Ferris

, Headrick

. Salmonella in dairy operations in the United States: Prevalence and antimicrobial drug susceptibility. J Food Prot, 2005; 68:696–702.

Callaway

, Edrington

, Rychlik

, Genovese

, Poole

, Jung

, Bischoff

, Anderson

, Nisbet

. Ionophores: Their use as ruminant growth promotants and impact on food safety. Curr Issues Intest Microbiol, 2003; 4:43–51.

[CLSI] Clinical Laboratory Standards Institute. Methods for Dilution Antimicrobial Susceptibility Tests for Bacteria That Grow Aerobically; Approved Standard, 8 ^thCLSI Document M07-A8Wayne, PA: CLSI, 2008.

Cummings

, Warnick

, Alexander

, Cripps

, McDonough

, Nydam

, Reed

. The incidence of salmonellosis among dairy herds in the northeastern United States. J Dairy Sci, 2009; 92:3766–3774.

Dohoo

, Martin

, Stryhn

. Veterinary Epidemiologic Research, 2ndCharlottetown, Prince Edward IslandCanada: VER, Inc., 2009.

Edrington

, Looper

, Duke

, Callaway

, Genovese

, Anderson

, Nisbet

. Effect of ionophore supplementation on the incidence of Escherichia coli O157:H7 and Salmonella and antimicrobial susceptibility of fecal coliforms in Stocker cattle. Foodborne Pathog Dis, 2006; 3:284–291.

Edrington

, Callaway

, Bischoff

, Genovese

, Elder

, Anderson

, Nisbet

. Effect of feeding the ionophores monensin and laidlomycin propionate and the antimicrobial bambermycin to sheep experimentally infected with E. coli O157:H7 and Salmonella Typhimurium. J Anim Sci, 2003; 81:553–560.

10.

Edrington

, Callaway

, Varey

, Jung

, Bischoff

, Elder

, Anderson

, Kutter

, Brabban

, Nisbet

. Effects of the antibiotic ionophores monensin, lasalocid, laidlomycin propionate and bambermycin on Salmonella and E. coli O157:H7 in vitro. J Appl Microbiol, 2003; 94:207–213.

11.

[FDA] U.S. Food and Drug Administration. National Antimicrobial Resistance Monitoring System—Enteric Bacteria (NARMS): 2008 Executive Report. Rockville, MD: FDA, 2011.

12.

Fossler

, Wells

, Kaneene

, Ruegg

, Warnick

, Bender

, Eberly

, Godden

, Halbert

. Herd-level factors associated with isolation of Salmonella in a multi-state study of conventional and organic dairy farms I. Salmonella shedding in cows. Prev Vet Med, 2005; 70:257–277.

13.

Fossler

, Wells

, Kaneene

, Ruegg

, Warnick

, Bender

, Eberly

, Godden

, Halbert

. Herd-level factors associated with isolation of Salmonella in a multi-state study of conventional and organic dairy farms II. Salmonella shedding in calves. Prev Vet Med, 2005; 70:279–291.

14.

Gardner

, Willeberg

, Mousing

. Empirical and theoretical evidence for herd size as a risk factor for swine diseases. Anim Health Res Rev, 2007; 3:43–55.

15.

Goff

. Pathophysiology of calcium and phosphorus disorders. Vet Clin North Am Food Anim Pract, 2000; 16:319–37, vii.

16.

Habing

, Lo

Y-J

, Kaneene

. Changes in the antimicrobial resistance profiles of Salmonella isolated from the same Michigan dairy farms in 2000 and 2009. Food Res Int, 2012; 45:919–924.

17.

Huston

, Wittum

, Love

, Keen

. Prevalence of fecal shedding of Salmonella spp in dairy herds. J Am Vet Med Assoc, 2002; 220:645–649.

18.

Kabagambe

, Wells

, Garber

, Salman

, Wagner

, Fedorka-Cray

. Risk factors for fecal shedding of Salmonella in 91 US dairy herds in 1996. Prev Vet Med, 2000; 43:177–194.

19.

Kaneene

, Warnick

, Bolin

, Erskine

, May

, Miller

. Changes in multidrug resistance of enteric bacteria following an intervention to reduce antimicrobial resistance in dairy calves. J Clin Microbiol, 2009; 47:4109–4112.

20.

Khachatryan

, Besser

, Hancock

, Call

. Use of a non-medicated dietary supplement correlates with increased prevalence of streptomycin-sulfa-tetracycline-resistant Escherichia coli on a dairy farm. Appl Environ Microb, 2006; 72:4583–4588.

21.

Maragakis

, Perencevich

, Cosgrove

. Clinical and economic burden of antimicrobial resistance. Expert Rev Anti Infect Ther, 2008; 6:751–763.

22.

Nicholson

, Groves

, Chambers

. Pathogen survival during livestock manure storage and following land application. Bioresource Technol, 2005; 96:135–143.

23.

Ray

, Warnick

, Mitchell

, Kaneene

, Ruegg

, Wells

, Fossler

, Halbert

, May

. Prevalence of antimicrobial resistance among Salmonella on midwest and northeast USA dairy farms. Prev Vet Med, 2007; 79:204–223.

24.

Ray

, Warnick

, Mitchell

, Kaneene

, Ruegg

, Wells

, Fossler

, Halbert

, May

. Antimicrobial susceptibility of Salmonella from organic and conventional dairy farms. J Dairy Sci, 2006; 89:2038–2050.

25.

Ruzante

, Lombard

, Wagner

, Fossler

, Karns

, Van Kessel

, Gardner

. Factors associated with Salmonella presence in environmental samples and bulk tank milk from US dairies. Zoonoses Public Health, 2010; 57:e217–e225.

26.

Scallan

. Foodborne illness acquired in the United States—Major pathogens. Emerg Infect Dis, 2011; 17:7–15.

27.

Toth

, Aceto

, Rankin

, Dou

. Survival characteristics of Salmonella enterica serovar Newport in the dairy farm environment. J Dairy Sci, 2011; 94:5238–5246.

28.

[USDA] U.S. Department of Agriculture. Dairy 2007, Part I: Reference of Dairy Cattle Health and Management Practices in the United States, 2007. Fort Collins, CO: USDA APHIS VS CEAH, 2007.

29.

[USDA] U.S. Department of Agriculture. Salmonella, Listeria, and Campylobacter on U.S. Dairy Operations, 1996–2007. Fort Collins, CO: USDA APHIS VS CEAH, 2010.

30.

[USDA] U.S. Department of Agriculture. Technical Bulletin: Clostridium difficile on U.S. Dairy Operations. Fort Collins, CO: USDA APHIS VS CEAH, 2011.

31.

Warnick

, Crofton

, Pelzer

, Hawkins

. Risk factors for clinical salmonellosis in Virginia, USA cattle herds. Prev Vet Med, 2001; 49:259–275.

32.

Wedel

, Bender

, Leano

, Boxrud

, Hedberg

, Smith

. Antimicrobial-drug susceptibility of human and animal Salmonella typhimurium, Minnesota, 1997–2003. Emerg Infect Dis, 2005; 11:1899–1906.

33.

Wells

, Fedorka-Cray

, Dargatz

, Ferris

, Green

. Fecal shedding of Salmonella spp. by dairy cows on farm and at cull cow markets. J Food Prot, 2001; 64:3–11.

34.

Zhao

, Qaiyumi

, Friedman

, Singh

, Foley

, White

, McDermott

, Donkar

, Bolin

, Munro

, Baron

, Walker

. Characterization of Salmonella enterica serotype Newport isolated from humans and food animals. J Clin Microbiol, 2003; 41:5366–5371.