Distribution of Multidrug-Resistant Human Isolatesof MDR-ACSSuT Salmonella Typhimurium and

MDR-AmpC Salmonella Newport in the United States,2003–2005

Sharon K. Greene,1,2,* Andrew M. Stuart,2 Felicita M. Medalla,2 Jean M. Whichard,2

Robert M. Hoekstra,2 and Tom M. Chiller2

Abstract

Purpose: Multidrug-resistant (MDR) Salmonella strains are associated with excess bloodstream infections,hospitalizations, and deaths compared with pansusceptible strains. Bovine products are sometimes asource of MDR Salmonella. To generate hypotheses for regional differences in risk factors for humaninfection, we analyzed distributions of the two most prevalent MDR Salmonella phenotypes in the UnitedStates, 2003–2005: (i) MDR-ACSSuT (resistant to at least ampicillin, chloramphenicol, streptomycin, sul-fonamides, and tetracycline) Typhimurium; (ii) MDR-AmpC (resistant to at least ampicillin, chloram-phenicol, streptomycin, sulfonamides, tetracycline, amoxicillin=clavulanic acid, and ceftiofur, and withdecreased susceptibility to ceftriaxone) Newport.Materials and Methods: Participating public health laboratories in all states forwarded every 20th Sal-monella isolate from humans to the National Antimicrobial Resistance Monitoring System for EntericBacteria for antimicrobial susceptibility testing. Among the serotypes Typhimurium and Newport isolatessubmitted 2003–2005, pansusceptible, MDR-ACSSuT Typhimurium, and MDR-AmpC Newport wereidentified. Patterns of resistance, demographic factors, and cattle density were compared across regions.Results: Of 1195 serotype Typhimurium isolates, 289 (24%) were MDR-ACSSuT. There were no significantdifferences in region, age, or sex distribution for pansusceptible versus MDR-ACSSuT Typhimurium. Of612 serotype Newport isolates, 97 (16%) were MDR-AmpC, but the percentage of MDR-AmpC isolatesvaried significantly across regions: South 3%, Midwest 28%, West 32%, and Northeast 38% ( p < 0.0001).The South had the lowest percentage of MDR-AmpC Newport isolates and also the lowest density of milkcows. More Newport isolates were MDR-AmpC in the 10 states with the highest milk cow density com-pared with the remaining states. Overall, 22% of pansusceptible Newport isolates but only 7% of MDR-AmpC Newport isolates were from patients <2 years of age. For both serotypes, MDR phenotypes had lessseasonal variation than pansusceptible phenotypes.Conclusion: This is the first analysis of the distribution of clinically important MDR Salmonella isolates inthe United States. MDR-ACSSuT Typhimurium was evenly distributed across regions. However, MDR-AmpC Newport was less common in the South and in children <2 years of age. Information on individuals’exposures is needed to fully explain the observed patterns.

1Epidemic Intelligence Service, Office of Workforce and Career Development, Centers for Disease Control and Prevention, Atlanta,Georgia.

2Division of Foodborne, Bacterial, and Mycotic Diseases, National Center for Zoonotic, Vectorborne, and Enteric Diseases, Centersfor Disease Control and Prevention, Atlanta, Georgia.

*Current affiliation: Department of Ambulatory Care and Prevention, Harvard Medical School and Harvard Pilgrim Health Care,Boston, Massachusetts.

FOODBORNE PATHOGENS AND DISEASEVolume 5, Number 5, 2008ª Mary Ann Liebert, Inc.DOI: 10.1089=fpd.2008.0111

669

Introduction

Salmonella enterica causes an estimated1.4 million illnesses and 400 deaths annually

in the United States (Voetsch et al., 2004). Of33,348 laboratory-confirmed S. enterica infec-tions with known serotypes reported to theCenters for Disease Control and Prevention(CDC) in 2005, 6982 (21%) were caused by se-rotype Typhimurium and 3295 (10%) werecaused by serotype Newport (CDC, 2006).These were, respectively, the first and thirdmost common serotypes in 2005 among culture-confirmed human salmonellosis cases.

Multidrug-resistant (MDR) Salmonella strainsassociated with resistance to clinically impor-tant drug classes, like cephalosporins and qui-nolones, are associated with excess bloodstreaminfections, hospitalizations, and death com-pared with pansusceptible strains (Helms et al.,2004; Varma et al., 2005; Varma et al., 2006). Tomonitor antimicrobial resistance among food-borne bacteria isolated from humans, the CDCarm of the National Antimicrobial ResistanceMonitoring System (NARMS) for Enteric Bac-teria was launched in 1996 in 14 sites (CDC,1996) (http:==www.cdc.gov=narms=). The pro-gram became nationwide in 2003 (CDC, 2007).

The emergence of S. enterica serotype Typhi-murium exhibiting the pentaresistant profileACSSuT (resistant to at least ampicillin, chlor-amphenicol, streptomycin, sulfonamides, andtetracycline) has prompted global concern (Suet al., 2004; Helms et al., 2005). Many Typhimur-ium strains belong to definitive phage type (DT)104, which emerged in the United States in the1990s (Glynn et al., 1998; Threlfall et al., 2000).In Salmonella serotype Typhimurium DT104,ACSSuT resistance is chromosomally mediated.Before NARMS began surveillance in 1996,MDR Typhimurium isolates were infrequentlyreported (Riley et al., 1984). In 1996, 34% of Ty-phimurium isolates from humans in NARMSwere ACSSuT, as were 23% of Typhimuriumisolates in 2004 (CDC, 2007).

Like MDR-ACSSuT Typhimurium, MDR-AmpC S. enterica serotype Newport strains areresistant to at least ampicillin, chloramphenicol,streptomycin, sulfonamides, and tetracycline.In addition, MDR-AmpC Newport isolates areresistant to amoxicillin=clavulanic acid and cef-

tiofur, and exhibit decreased susceptibility toceftriaxone (defined here as a minimum inhibi-tory concentration [MIC] $2mg=mL); ceftriax-one is the only antimicrobial recommended inthe United States for treating severe pediatricsalmonellosis (Zhao et al., 2003). MDR-AmpCresistance is conferred by a plasmid. In 1996, noNewport isolates in NARMS were MDR-AmpC,compared with 15% of isolates in 2004, a statis-tically significant increase (CDC, 2007).

Contact with cattle and consumption of bo-vine products, including commercial groundbeef and unpasteurized milk, are sources forMDR-ACSSuT Typhimurium (Dechet et al.,2006) and MDR-AmpC Newport (Gupta et al.,2003; Poppe et al., 2006; Varma et al., 2006; Karonet al., 2007). In addition, cattle density was spa-tially associated with the incidence of anotherhuman foodborne illness, verocytotoxigenicEscherichia coli, within Ontario, Canada (Michelet al., 1999).

Identifying regional differences in the distri-bution of resistant Salmonella infections mayindicate a heterogeneous distribution of riskfactors for this generally foodborne infectionand suggest interventions. Our objective was todescribe the seasonal and geographic distribu-tion of the two most prevalent MDR Salmonellaphenotypes in the United States: MDR-ACSSuTTyphimuriumandMDR-AmpCNewport (CDC,2007). Ecologic relationships between these in-fections and cattle density were also explored.

Materials and Methods

Isolate collection and testing

Public health laboratories forwarded every20th non-Typhi Salmonella isolate to CDCNARMS for antimicrobial susceptibility testing.Information on the patient’s age, sex, state ofresidence, and specimen collection date wassubmitted with each isolate. Some participatinglaboratories did not distinguish between sero-type Typhimurium variant O:5�=Copenhagenand serotype Typhimurium, so such isolateswere classified as serotype Typhimurium. Toevaluate representativeness, we determined thepercentage submitted to NARMS of all serotypeNewport and Typhimurium isolates reported tothe National Salmonella Surveillance System(NSSS), 2003–2005 (CDC, 2006). This percentage

670 GREENE ET AL.

was compared across the four Census regions(available at http:==www.census.gov=geo=www=us_regdiv.pdf, accessed April 26, 2007).

Isolates were tested using broth microdilu-tion (Sensititre�; Trek Diagnostics, Westlake,OH) to determine the MIC value for each of16 agents: amikacin, ampicillin, amoxicillin=clavulanic acid, cefoxitin, ceftiofur, ceftriaxone,cephalothin (in year 2003 only), chloramphenicol,ciprofloxacin, gentamicin, kanamycin, nalidixicacid, streptomycin, sulfamethoxazole (in year2003)=sulfisoxazole (in years 2004–2005), tetra-cycline, and trimethoprim=sulfamethoxazole.Results for the year 2005 were current as of April2007, but should be considered preliminary.Antimicrobial susceptibility results were avail-able for 1195 (99%) of 1204 Typhimurium iso-lates and 612 (99%) of 617 Newport isolatessubmitted to NARMS, from 2003 to 2005.

Clinical and Laboratory Standards Institute(2007) guidelines were used to interpret MICswhen available, and MIC results were dichoto-mized as resistant or susceptible. Isolates withintermediate susceptibility were categorized assusceptible. Isolates susceptible to all testedantimicrobials were classified as pansuscep-tible. The following were excluded from thedichotomizations: serotype Typhimurium iso-lates neither pansusceptible nor MDR-ACSSuT,and serotype Newport isolates neither pansus-ceptible nor MDR-AmpC. Results for each Ty-phimurium isolate were then dichotomized asat least resistance-type MDR-ACSSuT or pan-susceptible, and results for each Newport isolatewere dichotomized as at least MDR-AmpC orpansusceptible.

Cattle density calculations

The numbers of all cattle and calves, milkcows, beef cows, and acres of farmland for eachstate were determined from the 2002 Census ofAgriculture (National Agricultural StatisticsService, U.S. Department of Agriculture; avail-able at http:==www.agcensus.usda.gov=, ac-cessed January 11, 2007). Milk cows and beef

cows are cows used for reproductive purposesin milk and beef production.

For each state, the number of cattle in 2002was divided by the number of acres of farmlandin 2002, producing a cattle density measure(Michel et al., 1999). The following three densi-ties were calculated for each state per 10,000farmland acres: all cattle and calves, milk cows,and beef cows.

Statistical analysis

The percentage of Typhimurium isolates thatwas MDR-ACSSuT and the percentage ofNewport isolates that was MDR-AmpC weredetermined, and differences by sex, age group,1

season of specimen collection,2 and cattle den-sity were explored across regions. Analyseswere restricted to the 48 contiguous states andto 2003–2005, a period when NARMS surveil-lance was nationwide. Pearson chi-square testswere used to determine whether the percent-ages of pansusceptible versus MDR isolateswere nonhomogeneous across regional and de-mographic categories. P-values were based onthe application of exact methods.

For increased precision, an additional analy-sis was conducted at the state level. We exploredwhether a higher percentage of human isolateswere MDR in the 10 states most heavily in-volved in cattle production as compared withthe 38 other states.

Analyses were conducted using Excel 2003(Microsoft, Redmond, WA) and SAS 9.1 (SASInstitute, Cary, NC).

Results

Representativeness of NARMS

From 2003 to 2005, 20,376 Typhimurium iso-lates were reported to NSSS for the contiguousUnited States, and 1204 (6%) were submitted toNARMS. A total of 10,543 Newport isolates werereported to NSSS, and 617 (6%) were submittedto NARMS. The variation in the percentage of allisolates submitted to NARMS across the four

1<2 years, 2–9 years, 10–19 years, 20–64 years, and >64 years.2Spring (March, April, May), Summer ( June, July, August), Autumn (September, October, November), and Winter (December,

January, February).

DISTRIBUTION OF MULTIDRUG-RESISTANT SALMONELLA 671

Census regions was within �1% (5–7% for Ty-phimurium and 5–6% for Newport).

Distribution of pansusceptible and MDR

Salmonella Typhimurium and Newport

Of 1195 Typhimurium isolates tested inNARMS, 2003–2005, 721 (60%) were pansus-ceptible, 289 (24%) were MDR-ACSSuT, and theremaining 185 (16%) were excluded from di-chotomizations as neither pansusceptible norMDR-ACSSuT. The majority (61%) of the 474Typhimurium isolates resistant to $1 antimi-crobial agent were MDR-ACSSuT. Of the 289MDR-ACSSuT Typhimurium isolates, 27 (9%)were resistant to the additional antimicrobialagents conferring the MDR-AmpC phenotype.Of 612 Newport isolates, 492 (80%) were pan-susceptible, 97 (16%) were MDR-AmpC, and theremaining 23 (4%) were excluded from dichot-omizations as neither pansusceptible nor MDR-AmpC. The majority (81%) of the 120 Newportisolates resistant to $1 antimicrobial agent hadthe MDR-AmpC phenotype.

There was no statistically significant het-erogeneity in the percentage of pansusceptibleor MDR-ACSSuT Typhimurium across regions

(Table 1; Fig. 1A). The percentage of Typhi-murium isolates that was pansusceptibleranged from 53% in the West to 65% in theMidwest. The percentage of Typhimuriumisolates that was MDR-ACSSuT ranged nar-rowly from 22% in the Midwest to 26% in theSouth.

In contrast, there were regional differences inMDR-AmpC Newport prevalence. In the South,there was a low percentage of MDR-AmpC inNewport isolates (Table 1; Fig. 1B). The per-centage of Newport isolates that was pansus-ceptible ranged broadly from 59% in theNortheast to 95% in the South. The percentageof Newport isolates that was MDR-AmpC var-ied substantially across the regions, rangingfrom 3% in the South to 38% in the Northeast(Table 1). There was statistically significantheterogeneity in the percentage of both pan-susceptible Newport ( p< 0.0001) and MDRNewport ( p< 0.0001) across regions.

Differences in age, sex, and time

The distribution of isolates across age groupswas similar for those with pansusceptible andMDR-ACSSuT Typhimurium infections (Table

Table 1. Pansusceptible and Multidrug-Resistant (MDR) Salmonella Isolates, 2003–2005, and Cattle

Densities in 2002, by Census Region

Northeast Midwest South West Total

S. Typhimurium All isolates 212 299 448 236 1195Pansusceptible Count 128 194 273 126 721

Percentage 60.4 64.9 60.9 53.4 60.395% CI 53.8, 67.0 59.5, 70.3 56.4, 65.5 47.0, 59.8 57.6, 63.1

MDR-ACSSuT Count 48 67 115 59 289Percentage 22.6 22.4 25.7 25.0 24.295% CI 17.0, 28.3 17.7, 27.1 21.6, 29.7 19.5, 30.5 21.8, 26.6

S. Newport All isolates 88 93 342 89 612Pansusceptible Count 52 60 326 54 492

Percentage 59.1 64.5 95.3 60.7 80.495% CI 48.8, 69.4 54.8, 74.2 93.1, 97.6* 50.5, 70.8 77.2, 83.5

MDR-AmpC Count 33 26 10 28 97Percentage 37.5 28.0 2.9 31.5 15.895% CI 27.4, 47.6 18.8, 37.1 1.1, 4.7* 21.8, 41.1 13.0, 18.7

Cattle densities (number of All cattle and calves 181 104 127 69 102cattle per 1000 Milk cows 75 9 4 11 10acres of farmland) Beef cows 17 31 56 22 36

*Denotes statistical significance at a¼ 0.05.MDR-ACSSuT, resistant to at least ampicillin, chloramphenicol, streptomycin, sulfonamides, and tetracycline; MDR-AmpC,

resistant to at least ampicillin, chloramphenicol, streptomycin, sulfonamides, tetracycline, amoxicillin=clavulanic acid, and ceftiofur,and with decreased susceptibility to ceftriaxone; CI, confidence interval.

672 GREENE ET AL.

19

19

15

23

20

2423

2 1

3821

4

43

8

9

5

3

1015

120 1211

26

10

17

18

55

32

4

63

18

16

70

33

61

60

17

63

23

17

37

44

13

30

20

3

No resistance

0.1 - 16.7

16.8 - 30.2

30.3 - 42.9

43.0 - 61.5

61.6 - 100.0

No isolates submitted

Percentage Typhimurium-ACSSuTS.

63

1

3

4

2

914

10

9

48

17

14

18

70

6

17

27

14

15

16

1

2

7

3

7

4

8

53

7

2

11

3

5

1

13 35

19

13

9

9

714

2

No resistance

0.1 - 16.7

16.8 - 30.2

30.3 - 42.9

43.0 - 61.5

61.5 - 100.0

No isolates submitted

Percentage Newport-MDRAmpCS.

A

B

FIG. 1. Percentage of isolates that were multidrug-resistant by state, 2003–2005. The number of isolates of eachserotype submitted by each state is superimposed on the states. (A) MDR-ACSSuT Typhimurium. (B) MDR-AmpCNewport.

DISTRIBUTION OF MULTIDRUG-RESISTANT SALMONELLA 673

2A), as were the median ages (17 years vs. 19years). Of 665 pansusceptible and 264 MDR-ACSSuT Typhimurium isolates for which in-formation on patient sex was available, 315(47%) and 128 (49%), respectively, were fromfemales. There were no significant differencesin the sex distribution between pansusceptibleand MDR-ACSSuT Typhimurium (Table 2B).

Seasonality in pansusceptible infections wasapparent, as 65% of pansusceptible Typhimur-ium occurred in the summer and autumn. Incontrast, a lower percentage (52%) of MDR-ACSSuT Typhimurium occurred in these twoseasons. Significant differences in the seasonaldistribution between pansusceptible and MDR-ACSSuT Typhimurium occurred overall andwithin regions: in the Midwest and South, atleast one-third of MDR-ACSSuT Typhimuriuminfections occurred in winter (Table 2C).

In contrast with S. Typhimurium, there was asignificant overall difference in the age distri-bution between pansusceptible Newport andMDR-AmpC Newport. The median ages ofthose with pansusceptible and MDR-AmpCNewport infections were 22 years and 28 years,respectively. The sample size within each regionwas not large enough to detect intraregion agedifferences (Table 3A). Only 7% of MDR-AmpCNewport isolates were from patients<2 years ofage, compared with 22% of pansusceptibleNewport isolates.

Of 438 pansusceptible and 91 MDR-AmpCNewport isolates, 246 (56%) and 49 (54%), re-spectively, were from females. Of patients <2years of age with pansusceptible Newport in-fections, only 46% were female. There were nosignificant differences in the sex distributionbetween pansusceptible Newport and MDR-AmpC Newport (Table 3B).

This serotype also exhibited a more pro-nounced seasonality in pansusceptible thanMDR infections, since 79% of pansusceptibleNewport and only 62% of MDR-AmpC New-port infections occurred in summer and au-tumn. In the South, most MDR-AmpC Newportinfections were in autumn and winter, and noneoccurred in summer (Table 3C).

Cattle density and distribution of MDR isolates

Among the four regions, the Northeast had thehighest all cattle and calves and milk cow den-sities (Table 1); the Northeast also had the high-est percentage of MDR-AmpC Newport isolates.The South had the lowest milk cow density andthe highest beef cow density (Table 1); the Southalso had the lowest percentage of MDR-AmpCNewport isolates and the highest percentage ofMDR-ACSSuT Typhimurium isolates.

In the 10 states with the highest beef cowdensity,3 31% of Typhimurium isolates wereMDR-ACSSuT, compared with 22% in theremaining 38 states ( p¼ 0.002) (Table 4A).Conversely, the percentage of MDR-AmpCNewport in the 10 states with the highest beefcow density was only 3% compared with 21% inthe remaining states ( p< 0.0001) (Table 4B).

Significantly more Newport isolates wereMDR-AmpC in the 10 states with the highest allcattle and calves density4 ( p¼ 0.0001) and par-ticularly the 10 states with the highest milk cowdensity,5 ( p< 0.0001) compared with that in therespective groupings of the remaining states.

Discussion

This is the first description of regional dif-ferences in MDR Salmonella from a random,representative, national sample of humanclinical isolates collected in the United States.MDR-AmpC Newport had a 10-fold lowerpercentage of isolates in the South comparedwith the other three regions. Our exploratoryanalysis found a higher percentage of MDR-ACSSuT Typhimurium isolates in states withthe highest beef cow density and a higher per-centage of MDR-AmpC Newport isolates instates with the highest milk cow density. Ad-ditional studies to guide control efforts shoulddetermine the relationship between these pat-terns and bovine-related risk factors at the in-dividual level.

A high percentage of serotype Newportwas pansusceptible in the South, which maybe related to environmental and wild animalreservoirs (Srikantiah et al., 2004; Varma et al.,

3Florida, Tennessee, Alabama, Kentucky, Virginia, Missouri, Arkansas, Louisiana, Oklahoma, and Georgia.4Vermont, Wisconsin, Pennsylvania, Tennessee, California, New York, Virginia, Kentucky, Idaho, and Florida.5Vermont, New York, Wisconsin, Pennsylvania, Connecticut, California, Massachusetts, New Hampshire, Maryland, and Idaho.

674 GREENE ET AL.

2006; Greene et al., 2008). Wild animals have lessexposure to antimicrobial agents than domesticanimals, and bacteria in environmental andwild animal reservoirs experience less evolu-

tionary pressure to develop resistance. In addi-tion, the absolute number of MDR-AmpCNewport infections identified in the South waslow compared with the other regions, even

Table 2. Pansusceptible Serotype Typhimurium (Susc) and MDR-ACSSuT Typhimurium (MDR) Infections,

by Census Regiona

A.

Age < 2 (%) Age 2–9 (%) Age 10–19 (%) Age 20–64 (%) Age $65 (%) Exact p-value

Northeast Susc (n¼ 121) 24 (19.8) 29 (24.0) 12 (9.9) 47 (38.8) 9 (7.4) 0.08MDR (n¼ 47) 9 (19.2) 9 (19.2) 0 (0) 21 (44.7) 8 (17.0)

Midwest Susc (n¼ 176) 21 (11.9) 36 (20.5) 18 (10.2) 73 (41.5) 28 (15.9) 0.28MDR (n¼ 64) 12 (18.8) 13 (20.3) 6 (9.4) 29 (45.3) 4 (6.3)

South Susc (n¼ 244) 57 (23.4) 65 (26.6) 17 (7.0) 75 (30.7) 30 (12.3) 0.88MDR (n¼ 96) 20 (20.8) 26 (27.1) 7 (7.3) 34 (35.4) 9 (9.4)

West Susc (n¼ 118) 8 (6.8) 38 (32.2) 15 (12.7) 47 (39.8) 10 (8.5) 0.50MDR (n¼ 55) 5 (9.1) 14 (25.5) 11 (20.0) 23 (41.8) 2 (3.6)

Total Susc (n¼ 659) 110 (16.7) 168 (25.5) 62 (9.4) 242 (36.7) 77 (11.7) 0.62MDR (n¼ 262) 46 (17.6) 62 (23.7) 24 (9.2) 107 (40.8) 23 (8.8)

B.

Female (%) Exact p-value

Northeast Susc (n¼ 125) 58 (46.4) 0.86MDR (n¼ 47) 23 (48.9)

Midwest Susc (n¼ 180) 91 (50.6) 0.14MDR (n¼ 64) 25 (39.1)

South Susc (n¼ 241) 112 (46.5) 0.26MDR (n¼ 95) 48 (50.5)

West Susc (n¼ 119) 54 (45.4) 0.55MDR (n¼ 58) 32 (55.2)

Total Susc (n¼ 665) 315 (47.4) 0.77MDR (n¼ 264) 128 (48.5)

C.

Spring (%) Summer (%) Autumn (%) Winter (%) Exact p-value

Northeast Susc (n¼ 128) 31 (24.2) 50 (39.1) 30 (23.4) 17 (13.3) 0.29MDR (n¼ 48) 7 (14.6) 16 (33.3) 16 (33.3) 9 (18.8)

Midwest Susc (n¼ 194) 37 (19.1) 82 (42.3) 42 (21.6) 33 (17.0) 0.0002*MDR (n¼ 67) 12 (17.9) 12 (17.9) 17 (25.4) 26 (38.8)

South Susc (n¼ 273) 52 (19.1) 99 (36.3) 90 (33.0) 32 (11.7) < 0.0001*MDR (n¼ 115) 23 (20.0) 28 (24.4) 26 (22.6) 38 (33.0)

West Susc (n¼ 126) 33 (26.2) 41 (32.5) 35 (27.8) 17 (13.5) 0.53MDR (n¼ 59) 13 (22.0) 24 (40.7) 12 (20.3) 10 (17.0)

Total Susc (n¼ 721) 153 (21.2) 272 (37.7) 197 (27.3) 99 (13.7) <0.0001*MDR (n¼ 289) 55 (19.0) 80 (27.7) 71 (24.6) 83 (28.7)

aNortheast (Connecticut, Maine, Massachusetts, New Hampshire, New Jersey, New York, Pennsylvania, Rhode Island, Vermont)2002 population: 54.2 million.

Midwest (Illinois, Indiana, Iowa, Kansas, Michigan, Minnesota, Missouri, Nebraska, North Dakota, Ohio, South Dakota, Wisconsin)2002 population: 65.1 million.

South (Alabama, Arkansas, Delaware, Florida, Georgia, Kentucky, Louisiana, Maryland, Mississippi, North Carolina, Oklahoma,South Carolina, Tennessee, Texas, Virginia, West Virginia) 2002 population: 102.7 million.

West (Arizona, California, Colorado, Idaho, Montana, Nevada, New Mexico, Oregon, Utah, Washington, Wyoming) 2002population: 63.7 million.

*Denotes statistical significance at a¼ 0.05.P-values are shown for Pearson chi-square test. Comparisons by (A) age group, (B) sex, and (C) season.MDR-ACSSuT, resistant to at least ampicillin, chloramphenicol, streptomycin, sulfonamides, and tetracycline.

DISTRIBUTION OF MULTIDRUG-RESISTANT SALMONELLA 675

though the South had the largest population.The large denominator (342 Newport, of which326 were pansusceptible) and the small nu-merator (10 MDR-AmpC Newport) both con-tributed to the low percentage (3%) of MDR-AmpC Newport isolates in the South.

Pansusceptible Typhimurium, MDR-ACSSuTTyphimurium, and pansusceptible Newportwere all frequently isolated from patients <2

years of age. Infants, particularly those who arenot breastfed, are at increased risk for Salmonellainfection (CDC, 2001; Jones et al., 2006). How-ever, only six patients <2 years of age nation-wide, all in the Northeast, were identified ashaving MDR-AmpC Newport infection duringthe study period. This may indicate that youngchildren are rarely exposed to sources of MDR-AmpC Newport. Indeed, a 2002 population

Table 3. Pansusceptible Serotype Newport (Susc) and MDR-AmpC Newport (MDR) Infections,

by Census Region

A.

Age < 2 (%) Age 2–9 (%) Age 10–19 (%) Age 20–64 (%) Age $65 (%) Exact p-value

Northeast Susc (n¼ 50) 8 (16.0) 8 (16.0) 4 (8.0) 22 (44.0) 8 (16.0) 0.70MDR (n¼ 32) 6 (18.8) 4 (12.5) 4 (12.5) 16 (50.0) 2 (6.3)

Midwest Susc (n¼ 56) 4 (7.1) 7 (12.5) 4 (7.1) 27 (48.2) 14 (25.0) 0.31MDR (n¼ 25) 0 (0) 4 (16.0) 4 (16.0) 14 (56.0) 3 (12.0)

South Susc (n¼ 290) 83 (28.6) 59 (20.3) 25 (8.6) 84 (29.0) 39 (13.5) 0.29MDR (n¼ 8) 0 (0) 2 (25.0) 0 (0) 4 (50.0) 2 (25.0)

West Susc (n¼ 51) 4 (7.8) 6 (11.8) 5 (9.8) 31 (60.8) 5 (9.8) 0.36MDR (n¼ 25) 0 (0) 4 (16.0) 4 (16.0) 12 (48.0) 5 (20.0)

Total Susc (n¼ 447) 99 (22.2) 80 (17.9) 38 (8.5) 164 (36.7) 66 (14.8) 0.004*MDR (n¼ 90) 6 (6.7) 14 (15.6) 12 (13.3) 46 (51.1) 12 (13.3)

B.

Female (%) Exact p-value

Northeast Susc (n¼ 49) 28 (57.1) 0.66MDR (n¼ 33) 17 (51.5)

Midwest Susc (n¼ 56) 39 (69.6) 0.13MDR (n¼ 24) 12 (50.0)

South Susc (n¼ 284) 151 (53.2) 0.29MDR (n¼ 8) 6 (75.0)

West Susc (n¼ 49) 28 (57.1) 0.81MDR (n¼ 26) 14 (53.9)

Total Susc (n¼ 438) 246 (56.1) 0.73MDR (n¼ 91) 49 (53.8)

C.

Spring (%) Summer (%) Autumn (%) Winter (%) Exact p-value

Northeast Susc (n¼ 52) 9 (17.3) 21 (40.4) 18 (34.6) 4 (7.7) 0.99MDR (n¼ 33) 6 (18.2) 14 (42.4) 10 (30.3) 3 (9.1)

Midwest Susc (n¼ 60) 11 (18.3) 24 (40.0) 16 (26.7) 9 (15.0) 0.36MDR (n¼ 26) 7 (26.9) 6 (23.1) 10 (38.5) 3 (11.5)

South Susc (n¼ 326) 27 (8.3) 140 (42.9) 130 (39.9) 29 (8.9) 0.003*MDR (n¼ 10) 2 (20.0) 0 (0) 4 (40.0) 4 (40.0)

West Susc (n¼ 54) 7 (13.0) 24 (44.4) 15 (27.8) 8 (14.8) 0.53MDR (n¼ 28) 7 (25.0) 10 (35.7) 6 (21.4) 5 (17.9)

Total Susc (n¼ 492) 54 (11.0) 209 (42.5) 179 (36.4) 50 (10.2) 0.003*MDR (n¼ 97) 22 (22.7) 30 (30.9) 30 (30.9) 15 (15.5)

*Denotes statistical significance at a¼ 0.05.P-values are shown for Pearson chi-square test. Comparisons by (A) age group, (B) sex, and (C) season.MDR-AmpC, resistant to at least ampicillin, chloramphenicol, streptomycin, sulfonamides, tetracycline, amoxicillin=clavulanic acid,

and ceftiofur, and with decreased susceptibility to ceftriaxone.

676 GREENE ET AL.

survey showed the percentage of children <2years who ate ground beef in the past 7 days(33%) was much less than that of the overallpopulation (71%) (CDC, 2004).

The prominent seasonality of Salmonella in-fections, with increased infections in warmermonths, is well documented (CDC, 2001).However, seasonal differences between pan-susceptible and MDR infections have not pre-viously been reported. For both serotypes,differences were most apparent in the South,and pansusceptible isolates exhibited a moresharply pronounced seasonal pattern than MDRisolates. This seasonality suggests different riskfactors for pansusceptible and MDR human in-fections across seasons—for example, variationin the prevalence of carriage in food animals,cooking behaviors, or consumption patterns.

The use of antimicrobial agents in cattle hasbeen associated with the emergence of antimi-crobial-resistant nontyphoidal Salmonella strainsand transmission of these strains to humans(Holmberg et al., 1984; Spika et al., 1987; Anguloet al., 2000; Fey et al., 2000; Threlfall et al., 2000).This is consistent with our findings of a higherpercentage of MDR-ACSSuT Typhimurium

isolates in states with the highest beef cowdensity and a higher percentage of MDR-AmpCNewport isolates in states with the highest milkcow density. One of the states with the highestmilk cow density, Wisconsin, reported a recentincrease in MDR Newport (Karon et al., 2007).

Epidemiologic studies have demonstratedthat contact with cattle and consumption ofbovine food products are sources of MDR-AmpC Newport infection. These infections havebeen associated with exposure to cattle, unpas-teurized milk (Karon et al., 2007), and farms(Gupta et al., 2003; Karon et al., 2007), as well asuncooked ground beef consumption (Varma etal., 2006). A Canadian study noted an increase inMDR-AmpC Newport infections in dairy cattleduring 1993–2002 (Poppe et al., 2006). From2000 to 2005, Newport was among the fourSalmonella serotypes the Food Safety and In-spection Service most frequently isolated fromground beef (U.S. Department of Agriculture,2006). A transparent national data-sharing net-work for resistance phenotypes and molecularcharacteristics of human and food animal Sal-monella isolates would strengthen foodbornedisease surveillance.

Table 4. Percentage of Isolates That Were Multidrug-Resistant in the 10 States Most Heavily

Involved in Each of the Three Different Measures of Cattle Production, as Compared to the 38 Other

Contiguous States

A. Serotype Typhimurium

All cattle and calves density Milk cow density Beef cow density

PercentageMDR-ACSSuT

Exactp-value

PercentageMDR-ACSSuT

Exactp-value

PercentageMDR-ACSSuT

Exactp-value

Top 10 states 27.3 0.11 22.0 0.27 31.1 0.002*Bottom 38 states 22.8 25.1 21.8

B. Serotype Newport

All cattle and calves density Milk cow density Beef cow density

PercentageMDR-AmpC

Exactp-value

PercentageMDR-AmpC

Exactp-value

PercentageMDR-AmpC

Exactp-value

Top 10 states 25.3 0.0001* 35.1 <0.0001* 3.4 <0.0001*Bottom 38 states 12.2 9.5 20.9

*Denotes statistical significance at a¼ 0.05.P-values are shown for Pearson chi-square test with one degree of freedom.MDR-ACSSuT, resistant to at least ampicillin, chloramphenicol, streptomycin, sulfonamides, and tetracycline; MDR-AmpC,

resistant to at least ampicillin, chloramphenicol, streptomycin, sulfonamides, tetracycline, amoxicillin=clavulanic acid, and ceftiofur,and with decreased susceptibility to ceftriaxone.

DISTRIBUTION OF MULTIDRUG-RESISTANT SALMONELLA 677

This study is subject to several limitations.First, isolate submission was affected by healthcare–seeking behavior and did not fully reflectthe underlying population’s age structure.Therefore, it was not appropriate to calculateage-specific infection incidences. Second, riskfactor information for individuals is not col-lected in routine NARMS surveillance, so ob-served associations between resistant infectionsand cattle were ecological. Third, publisheddata on total numbers of cattle in the beef anddairy industries, by state, are unavailable.Counts of beef cows and milk cows includedcows used for reproductive purposes in beefand milk production, but calves, replacementheifers, steers, bulls, and cattle fed for theslaughter market were excluded. Finally, cattledistributions may correspond to opportunitiesfor on-farm exposures, but are an impreciseproxy for specific bovine-related risk factors,such as ground beef consumption. The locationswhere cattle are raised are not necessarily thesame locations as where bovine products areprocessed or consumed.

Regional bias in isolate submission toNARMS leading to differences in reported re-sistance levels was unlikely to have affected thisstudy. The percentage of all Newport and Ty-phimurium infections reported to NSSS forwhich isolates were submitted to NARMS wasnear 5% for all regions, consistent with the non-Typhi Salmonella isolate submission scheme.

This report suggests interesting hypothesesregarding local farm contact and meat con-sumption as risk factors. Additional studies areneeded to identify individual-level causes ofregional and demographic differences in MDRSalmonella and to determine the fraction ofMDR-ACSSuT Typhimurium and MDR-AmpCNewport infections that can be attributed tobovine products and other sources. State-levelground beef distribution and consumption data,as well as laboratory subtype data on isolatesfrom food-producing animals, are needed tofurther explore risk factors related to humaninfection. For example, MDR-AmpC Newporthas been isolated from swine and chickens(Zhao et al., 2003). Salmonella with ceftiofur re-sistance (a reliable marker for MDR-AmpC)have been isolated not only from ground beef,but also less frequently from chicken breasts,

pork chops, and ground turkey purchased atgrocery stores (Zhao et al., 2006). Future studiesshould compare the distribution of resistantstrains among isolates from humans and food-producing animals to the density of food ani-mals other than cattle.

Because MDR-AmpC resistance is conferredby a plasmid, this phenotype may spread easily.The MDR-AmpC phenotype has been identifiedin numerous other Salmonella serotypes besidesNewport (Medalla et al., 2006). Similar concernsmay apply for Salmonella serotype Typhimur-ium DT104, given the mobilization potential ofthe genomic island housing the resistance genes(Doublet et al., 2005). Efforts to analyze thespread of resistance genes and to control furtherdistribution are needed. For example, an ex-ploration of factors that might facilitate thedissemination of mobile resistance elements iswarranted.

Acknowledgments

The authors thank Bernadette Hartman (En-teric Diseases Epidemiology Branch, CDC) forassistance in interpreting cattle production dataand the NARMS participants for contributingisolates and data.

The findings and conclusions in this reportare those of the authors and do not necessarilyrepresent the views of the Centers for DiseaseControl and Prevention.

References

Angulo, F.J., K.R. Johnson, R.V. Tauxe, and M.L. Cohen.2000. Origins and consequences of antimicrobial-resistant nontyphoidal Salmonella: implications for theuse of fluoroquinolones in food animals. Microb. Drug.Resist. 6:77–83.

CDC. 1996. Establishment of a national surveillance pro-gram for antimicrobial resistance in Salmonella. MMWRMorb. Mortal. Wkly. Rep. 45:110–111.

CDC. 2001. An atlas of Salmonella in the United States:serotype-specific surveillance, 1968–1998. U.S. Depart-ment of Health and Human Services, Centers for DiseaseControl and Prevention, Atlanta, GA.

CDC. 2004. Foodborne Diseases Active SurveillanceNetwork (FoodNet): Population Survey Atlas of Expo-sures, 2002. Centers for Disease Control and Preven-tion, Atlanta, GA. Available at: http:==www.cdc.gov=foodnet=surveys=pop=2002=2002Atlas.pdf (online).

CDC. 2006. Salmonella surveillance: annual summary,2005. US Department of Health and Human Services,

678 GREENE ET AL.

Centers for Disease Control and Prevention, Atlanta,GA.

CDC. 2007. National Antimicrobial Resistance MonitoringSystem for Enteric Bacteria (NARMS) 2004 human iso-lates final report. U.S. Department of Health and HumanServices, CDC, Atlanta, GA.

Clinical and Laboratory Standards Institute. 2007. Perfor-mance standards for antimicrobial susceptibility testing:seventeenth informational supplement. CLSI=NCCLSdocument M100-S17. Clinical and Laboratory StandardsInstitute, Wayne, PA.

Dechet, A.M., E. Scallan, K. Gensheimer, R. Hoekstra,J. Gunderman-King, J. Lockett, D. Wrigley, W. Chege,and J. Sobel. 2006. Outbreak of multidrug-resistant Sal-monella enterica serotype Typhimurium Definitive Type104 infection linked to commercial ground beef, north-eastern United States, 2003–2004. Clin. Infect. Dis.42:747–752.

Doublet, B., D. Boyd, M.R. Mulvey, and A. Cloeckaert.2005. The Salmonella genomic island 1 is an integrativemobilizable element. Mol. Microbiol. 55:1911–1924.

Fey, P.D., T.J. Safranek, M.E. Rupp, E.F. Dunne, E. Ribot,P.C. Iwen, P.A. Bradford, F.J. Angulo, and S.H. Hinrichs.2000. Ceftriaxone-resistant salmonella infection ac-quired by a child from cattle. N. Engl. J. Med. 342:1242–1249.

Glynn, M.K., C. Bopp, W. Dewitt, P. Dabney, M. Mokhtar,and F.J. Angulo. 1998. Emergence of multidrug-resistantSalmonella enterica serotype Typhimurium DT104 infec-tions in the United States. N. Engl. J. Med. 338:1333–1338.

Gupta, A., J. Fontana, C. Crowe, B. Bolstorff, A. Stout, S.van Duyne, M.P. Hoekstra, J.M. Whichard, T.J. Barrett,and F.J. Angulo. 2003. Emergence of multidrug-resistantSalmonella enterica serotype Newport infections resistantto expanded-spectrum cephalosporins in the UnitedStates. J. Infect. Dis. 188:1707–1716.

Greene, S.K., E.R. Daly, E.A. Talbot, L.J. Demma, S.Holzbauer, N.J. Patel, T.A. Hill, M.O. Walderhaug, R.M.Hoekstra, M.F. Lynch, and J.A. Painter. 2008. Recurrentmultistate outbreak of Salmonella Newport associatedwith tomatoes from contaminated fields, 2005. Epide-miol. Infect. 136:157–165.

Helms, M., S. Ethelberg, and K. Molbak. 2005. Interna-tional Salmonella Typhimurium DT104 infections, 1992–2001. Emerg. Infect. Dis. 11:859–867.

Helms, M., J. Simonsen, and K. Molbak. 2004. Quinoloneresistance is associated with increased risk of invasiveillness or death during infection with Salmonella serotypeTyphimurium. J. Infect. Dis. 190:1652–1654.

Holmberg, S.D., M.T. Osterholm, K.A. Senger, and M.L.Cohen. 1984. Drug-resistant Salmonella from animals fedantimicrobials. N. Engl. J. Med. 311:617–622.

Jones, T.F., L.A. Ingram, K.E. Fullerton, R. Marcus, B.J.Anderson, P.V. McCarthy, D. Vugia, B. Shiferaw, N.Haubert, S. Wedel, and F.J. Angulo. 2006. A case-controlstudy of the epidemiology of sporadic Salmonella infec-tion in infants. Pediatrics 118:2380–2387.

Karon, A.E., J.R. Archer, M.J. Sotir, T.A. Monson, and J.J.Kazmierczak. 2007. Human multidrug-resistant Salmo-

nella Newport infections in a major dairy state, 2003–2005. Emerg. Infect. Dis. 13:1777–1780.

Medalla, F., J. Whichard, A. Stuart, K. Joyce, R.M. Hoekstra,and T.M. Chiller. 2006. Emergence of a multidrug-resistant phenotype, MDR-AmpC, in non-Typhi Salmo-nella, NARMS, 1996–2004 (poster presentation), 46thAnnual Interscience Conference on AntimicrobialAgents and Chemotherapy. San Francisco, CA.

Michel, P., J.B. Wilson, S.W. Martin, R.C. Clarke, S.A.McEwen, and C.L. Gyles. 1999. Temporal and geo-graphical distributions of reported cases of Escherichiacoli O157:H7 infection in Ontario. Epidemiol. Infect.122:193–200.

Poppe, C., L. Martin, A. Muckle, M. Archambault, S.McEwen, and E. Weir. 2006. Characterization of anti-microbial resistance of Salmonella Newport isolated fromanimals, the environment, and animal food products inCanada. Can. J. Vet. Res. 70:105–114.

Riley, L.W., M.L. Cohen, J.E. Seals, M.J. Blaser, K.A. Birk-ness, N.T. Hargrett, S.M. Martin, and R.A. Feldman.1984. Importance of host factors in human salmonellosiscaused by multiresistant strains of Salmonella. J. Infect.Dis. 149:878–883.

Spika, J.S., S.H. Waterman, G.W. Hoo, M.E. St Louis, R.E.Pacer, S.M. James, M.L. Bissett, L.W. Mayer, J.Y. Chiu, B.Hall, et al. 1987. Chloramphenicol-resistant SalmonellaNewport traced through hamburger to dairy farms: amajor persisting source of human salmonellosis in Ca-lifornia. N. Engl. J. Med. 316:565–570.

Srikantiah, P., J.C. Lay, S. Hand, J.A. Crump, J. Campbell,M.S. van Duyne, R. Bishop, R. Middendor, M. Currier,P.S. Mead, and K. Molbak. 2004. Salmonella enterica se-rotype Javiana infections associated with amphibiancontact, Mississippi, 2001. Epidemiol. Infect. 132:273–281.

Su, L.H., C.H. Chiu, C. Chu, and J.T. Ou. 2004. Anti-microbial resistance in nontyphoid Salmonella serotypes:a global challenge. Clin. Infect. Dis. 39:546–551.

Threlfall, E.J., L.R. Ward, J.A. Frost, and G.A. Willshaw.2000. The emergence and spread of antibiotic resistancein food-borne bacteria. Int. J. Food Microbiol. 62:1–5.

U.S. Department of Agriculture, Food Safety and Inspec-tion Service. 2006. Serotypes profile of Salmonella isolatesfrom meat and poultry products, January 1998 throughDecember 2005. U.S. Department of Agriculture,Washington, DC. Available at http:==www.fsis.usda.gov=Science=Serotypes_Profile_Salmonella_Isolates=index.asp (online).

Varma, J.K., K.D. Greene, J. Ovitt, T.J. Barrett, F. Medalla,and F.J. Angulo. 2005. Hospitalization and antimicrobialresistance in Salmonella outbreaks, 1984–2002. Emerg.Infect. Dis. 11:943–946.

Varma, J.K., R. Marcus, S.A. Stenzel, S.S. Hanna, S. Gett-ner, B.J. Anderson, T. Hayes, B. Shiferaw, T.L. Crume, K.Joyce, K.E. Fullerton, A.C. Voetsch, and F.J. Angulo.2006. Highly resistant Salmonella Newport-MDRAmpCtransmitted through the domestic US food supply: aFoodNet case-control study of sporadic SalmonellaNewport infections, 2002–2003. J. Infect. Dis. 194:222–230.

DISTRIBUTION OF MULTIDRUG-RESISTANT SALMONELLA 679

Voetsch, A.C., T.J. van Gilder, F.J. Angulo, M.M. Farley,S. Shallow, R. Marcus, P.R. Cieslak, V.C. Deneen, andR.V. Tauxe. 2004. FoodNet estimate of the burden of ill-ness caused by nontyphoidal Salmonella infections in theUnited States. Clin. Infect. Dis. 38 Suppl 3:S127–S134.

Zhao, S., P.F. McDermott, S. Friedman, J. Abbott, S. Ayers,A. Glenn, E. Hall-Robinson, S.K. Hubert, H. Harbottle,R.D. Walker, T.M. Chiller, and D.G. White. 2006. Anti-microbial resistance and genetic relatedness among Sal-monella from retail foods of animal origin: NARMS retailmeat surveillance. Foodborne Pathog Dis 3:106–117.

Zhao, S., S. Qaiyumi, S. Friedman, R. Singh, S.L. Foley,D.G. White, P.F. McDermott, T. Donkar, C. Bolin, S.

Munro, E.J. Baron, and R.D. Walker. 2003. Character-ization of Salmonella enterica serotype Newport isolatedfrom humans and food animals. J. Clin. Microbiol.41:5366–5371.

Address reprint requests to:Tom M. Chiller, M.D., M.P.H.

Centers for Disease Control and Prevention1600 Clifton Road, MS C-09

Atlanta, GA 30333

E-mail: TChiller@cdc.gov

680 GREENE ET AL.