Abstract

Background. Most human infections are caused by closely related serotypes within 1 species of Salmonella. Few data are available on differences in severity of disease among common serotypes.

Methods. We examined data from all cases of Salmonella infection in FoodNet states during 1996–2006. Data included serotype, specimen source, hospitalization, and outcome.

Results. Among 46,639 cases, 687 serotypes were identified. Overall, 41,624 isolates (89%) were from stool specimens, 2524 (5%) were from blood, and 1669 (4%) were from urine; 10,393 (22%) cases required hospitalization, and death occurred in 219 (0.5%). The case fatality rate for S. Newport (0.3%) was significantly lower than for Typhimurium (0.6%); Dublin (3.0%) was higher. With respect to invasive disease, 13 serotypes had a significantly higher proportion than Typhimurium (6%), including Enteritidis (7%), Heidelberg (13%), Choleraesuis (57%), and Dublin (64%); 13 serotypes were significantly less likely to be invasive. Twelve serotypes, including Enteritidis (21%) and Javiana (21%), were less likely to cause hospitalization than Typhimurium (24%); Choleraesuis (60%) was significantly more so.

Conclusions. Salmonella serotypes are closely related genetically yet differ significantly in their pathogenic potentials. Understanding the mechanisms responsible for this may be key to a more general understanding of the invasiveness of intestinal bacterial infections.

Salmonella infections are common and can result in protean clinical manifestations, ranging from an asymptomatic state to very severe disease. Non-Typhi salmonellae cause an estimated 1.4 million illnesses in the United States each year, resulting in an estimated 15,000 hospitalizations and >400 deaths [1]. More than 2500 serotypes of Salmonella have been identified, and they display a remarkable propensity to fill a variety of ecological niches. Most human infections are caused by one species (Salmonella enterica) that can be differentiated, on the basis of agglutination reactions, into closely related serotypes. Although Salmonella serotypes Typhi and Paratyphi are common pathogens in developing countries, they account for <1% of Salmonella infections in the United States [2], and most of those cases are associated with international travel. Certain uncommon serotypes (such as S. Choleraesuis and S. Dublin) are known to cause more severe disease than other non-Typhi salmonellae [3]. However, significant differences in the severity of disease among more common non-Typhi Salmonella serotypes are poorly understood and likely underappreciated.

Previous studies of the comparative virulence of various Salmonella serotypes have often been small [4–7], included only a limited number of serotypes [4–11], were performed in areas with substantially different distributions of predominant serotypes [4, 5], or used only the end point of isolation of Salmonella from blood [3–8]. In addition, most of these studies preceded changes in nomenclature, serotype distributions, antimicrobial resistance, virulence, and therapeutic advances that may substantially affect the applicability of their results in the current medical environment.

Understanding the relative risks among different Salmonella serotypes for invasive infection and poor outcomes has important implications for patients, medical providers, and the public health community. Salmonella serotyping is typically not performed at clinical laboratories but at state public health laboratories. It is therefore important that Salmonella isolates be submitted to public health laboratories and that serotyping results be reported promptly to providers. We analyzed data from a large population-based active surveillance program with active ascertainment of clinical outcomes to assess the current status of outcomes of Salmonella infections by serotype.

Methods

Salmonella surveillance data from FoodNet collected during 1996–2006 were included in the study. FoodNet is a collaborative project involving the Centers for Disease Control and Prevention's Emerging Infections Program, selected state health departments, the US Department of Agriculture Food Safety Inspection Service, and the US Food and Drug Administration. During 1996–2006, the FoodNet surveillance population increased from 14.2 million persons (5% of the US population) in 5 states to 44.9 million persons (15% of the US population) in 10 states. Population-based active surveillance for culture-confirmed Salmonella infections was performed at all clinical laboratories of participating sites [12].

Data collected included patient age, race, source and date of specimen collection, Salmonella serotype, hospitalization, and outcome (death or survival). For this study, S. Typhi and S. Paratyphi were not included. For comparing rates of clinical outcomes, S. Typhimurium (the most common Salmonella serotype in the United States) was used as the comparator. In calculating rates of death, cases for which the outcome was unknown were counted as alive. For purposes of analysis, Salmonella serogroups II-IV, which typically cause <1% of human disease, were grouped together. There were no isolates of serogroups V or VI. Serogroup I salmonellae are most commonly isolated from humans and other warm-blooded animals, and other serogroups are more typically associated with reptiles.

Invasive disease was defined as isolation of Salmonella from blood, cerebrospinal fluid (CSF), bone or joint fluid, or a sterile site not further specified. Urine, abscess, and wound cultures were not categorized as sterile sites because sufficient data to differentiate between invasive infection and contamination or colonization were not available. Salmonella isolates from urine were not collected during 1996–1998. Hospitalization was defined as hospitalization within 7 days after specimen collection. Death was defined as death within 7 days after Salmonella isolation if an outpatient or, for inpatients, failure to survive the hospitalization. Data were analyzed using SAS (version 9.1; SAS Institute). Data were collected as part of public health surveillance activities as determined by the Centers for Disease Control and Prevention and participating site institutional review boards.

Results

From 1996 through 2006, a total of 51,964 cases of non-Typhi Salmonella infection were identified by FoodNet. Of these, 5325 (10%) had incomplete serotyping information available, leaving 46,639 cases for analysis, including 687 different serotypes. The most commonly isolated serotypes were Typhimurium, Enteritidis, Newport, Heidelberg, and Javiana, which together accounted for 61% of all isolates (table 1).

Table 1.

Distribution of Salmonella serotypes with >50 cases identified during 1996–2006 and proportions of specified outcomes.

Table 1.

Distribution of Salmonella serotypes with >50 cases identified during 1996–2006 and proportions of specified outcomes.

Of all Salmonella cases, 41,624 (89%) had an isolate from a stool specimen, 2524 (5%) from blood, 102 (0.2%) from an abscess, 31 (0.06%) from bone or joint fluid, 19 (0.04%) from CSF, and 567 (1%) from unknown or unspecified sites. From 1999 through 2004, 1669 isolates (4%) were from urine. Overall, 10,393 cases (22%) required hospitalization (range, 17%–27% per year), 2524 (5%) caused invasive disease (range, 4%–6% per year), and 219 (0.5%) caused death (range, 0.3%–0.8% per year). Rates of adverse outcomes varied substantially by age and followed a J-shaped curve in each case (figure 1). Among persons ≥65 years old, the proportions of hospitalization (49%), invasive disease (12%), and death (3%) were significantly higher than those among persons <65 years of age (20%, 5%, and 0.3%, respectively).

Figure 1.

Proportion of cases of Salmonella infection requiring hospitalization, by age group. Similar J-shaped curves were observed for rates of invasive disease and death.

Figure 1.

Proportion of cases of Salmonella infection requiring hospitalization, by age group. Similar J-shaped curves were observed for rates of invasive disease and death.

The proportions of all Salmonella cases with isolates from various sources and rates of specific outcomes did not change substantially over the study period, except that the mean proportion of cases requiring hospitalization each year from 2003 through 2006 (26%) was higher than that during the years 1996 through 2002 (19%; rate ratio [RR], 1.35 [95% confidence interval {CI}, 1.30–1.39]). Overall rates of salmonellosis and specific outcomes did not differ substantially by sex. Compared with nonwhite persons, white persons had significantly lower rates of hospitalization (RR, 0.75 [95% CI, 0.78–0.81]), invasive disease (RR, 0.31 [95% CI, 0.33–0.36]), and death (RR, 0.51 [95% CI, 0.38–0.69]). Overall rates of hospitalization (RR, 1.16 [95% CI, 1.12–1.22]), invasive disease (RR, 1.48 [95% CI, 1.35–1.61]), and death (RR, 2.27 [95% CI, 1.70–3.02]) among persons with Salmonella infection were higher during winter months compared with other seasons.

Of 10,894 S. Typhimurium cases, 70 (0.6%) caused death. S. Dublin had a significantly higher case fatality rate (3%) than other serotypes (table 1). S. Newport had a significantly lower case fatality rate (0.3%) than S. Typhimurium; case fatality rates of other serotypes did not differ significantly from that of S. Typhimurium.

Thirteen serotypes had a significantly higher proportion of invasive disease than S. Typhimurium (6%), including Enteritidis (7%), Heidelberg (13%), Schwarzengrund (15%), Poona (17%), Choleraesuis (57%), and Dublin (64%); 13 other serotypes (the most common of which included Infantis, Javiana, Braenderup, Newport, and Muenchen) were significantly less likely to be isolated from sterile sites.

Cases of S. Choleraesuis (60%) and S. Dublin (67%) infection were more likely to cause hospitalization than all other serotypes. Overall, the mean duration of stay for persons hospitalized with Salmonella infection was 5.3 days (median, 3.0 days), with no substantial variation among the common serotypes.

The overall proportion of persons hospitalized was higher for those with Salmonella isolated from invasive sites (32%) than for those with stool isolates (20%). A similar difference was observed for rates of death (3.3% and 0.2%, respectively). These differences were observed consistently among common serotypes.

The mean age of persons infected with the 20 most common Salmonella serotypes ranged from 18 to 33 years. Dublin (mean age, 54 years) was the only one of the 50 most common serotypes for which the mean age was >40 years. Persons with Choleraesuis infection had a mean age of 45 years. Rates of hospitalization, invasive disease, and death did not differ significantly among patients <65 years of age, compared with those ≥65, for either Dublin or Choleraesuis.

Of 2512 isolates from blood, 598 (24%) were S. Typhimurium, 486 (19%) were S. Enteritidis, and 370 (15%) were S. Heidelberg. Of the 22 most common Salmonella serotypes, the proportion of isolates recovered from urine ranged from 2% to 6%, with the exception of Anatum (13%) and Mbandaka (12%).

Among 5783 Salmonella cases in infants <1 year of age, the most common serotypes were Typhimurium (21%), Newport (13%), Javiana (7%), Enteritidis (6%), Heidelberg (5%), Muenchen (5%), and Montevideo (4%). Of the infants, 23% were hospitalized, 6% had invasive disease, and 0.1% died. These proportions were significantly lower than those of adults ≥18 years of age (RR for hospitalization, 0.84 [95% CI, 0.80–0.89]; RR for invasive disease, 0.80 [95% CI, 0.71–0.90]; RR for death, 0.13 [95% CI 0.06]). Of infant deaths, 2 were associated with Enteritidis and 1 each with Newport, Mississippi, Muenster, and IV 16:Z4,Z32:-(formerly Chameleon). Overall, only 19 isolates of 11 serotypes were from CSF. Of these, 14 (74%) were from persons ≤1 year of age, of which 11 (58%) were from persons ≤2 months of age.

The annual rate of Salmonella infection due to all serotypes was 13.2 cases/100,000 population; the rate of hospitalization was 2.9, of bloodstream infection was 0.7, and of death was 0.06/ 100,000 population. The rate of Salmonella infection was highest among infants <1 year of age (120/100,000 population). The population rates of specific outcomes among different age groups followed J-shaped curves; rates of hospitalizations due to Salmonella were 27/100,000 population in infants, 1.6/100,000 population in persons aged 15–19 years, and 5.1/100,000 population in persons ≥65 years of age. Bloodstream infections (6.5, 0.3, and 1.2/100,000 population, respectively) and death (0.12, 0, and 0.3/100,000 population) followed a similar pattern.

Discussion

Not all salmonellae are created equal. The differences in the severity of disease among different serotypes are substantial. The present study analyzes a much larger number of Salmonella cases than any published to date and provides data on a number of different outcomes and a wide spectrum of serotypes. We found that case fatality rates among common Salmonella serotypes ranged over 100-fold and that the proportions of cases hospitalized varied from 14% to 67%. These data have important implications for better understanding the public health effect and pathogenicity of salmonellosis.

It has long been known that infections with S. Choleraesuis are more likely to be invasive and lead to poor outcomes than other non-Typhi serotypes [13, 14]. At the time when many of these studies were performed, however, S. Choleraesuis was among the half dozen most common serotypes identified [14–16]. This has changed dramatically, and now S. Choleraesuis accounts for only 0.1% of salmonellae isolated. Some of this apparent difference may be an artifact of changes in nomenclature, given that, before the adoption of S. enterica as a species designation, S. choleraesuis referred to both a species and a serotype [17]. Other Salmonella serotypes identified in various older studies as leading to death in a high proportion of cases have included Derby, Bareilly, Newport, Oranienburg, Montevideo, Enteritidis, Panama, and Anatum [9, 14–16], although results vary substantially among studies, most of which were limited by low numbers. Furthermore, few data are available on outcomes associated with the large majority of serotypes encountered in current practice. A study published in 1981 identified a rate of Salmonella bacteremia of 2.8 cases/million population [3], which is one-third of the population rate in this study. In addition, the overall population-based rate of Salmonella infection in our study was 37% higher. These differences likely reflect the high case-ascertainment rates and active identi cation of isolate source by FoodNet sites.

The absolute case fatality rates associated with salmonellosis have fallen steadily in successive studies [9, 14–16] and are now a tenth what they were in the 1940s. Not surprisingly, given the current era of improved management and therapy, overall case fatality rates in our study were low and varied over a relatively narrow range, with few significant differences among serotypes. Other markers of disease severity are probably more instructive in the setting of current standards of care.

The relative likelihoods of more severe outcomes are difficult to compare among past studies, because most were limited to assessing death or bacteremia and had few cases of infection with less common serotypes. We are not aware of any studies comparing rates of hospitalization among Salmonella serotypes, which is another measure of disease severity, albeit one influenced by local practice and evolving standards of care. Of note, rates of adverse outcomes were higher among nonwhite persons than among white persons. Although this might be due to decreased access to care or correlate with socioeconomic or educational factors, we did not have the data to assess this. We also found that rates of hospitalization were higher in later years during the study period. Additional studies including data on socioeconomic factors, clinical details, and comorbidities would be instructive in better understanding these observations.

The most common serotypes isolated from persons with salmonellosis have changed considerably over time. In 5 different studies over several decades, the lists of the 5 most common salmonellae never included all of the same serotypes, much less in the same order [3, 9, 14–16], and the current list is no exception. Although many past studies have identified the serotypes responsible for the largest proportion of deaths or bacteremia, such results are highly dependent on the prevalence of particular serotypes. Understanding the proportion of specific serotypes associated with particular outcomes is more useful, because such measures do not change as the relative prevalences of common serotypes shift.

Because Salmonella serotyping is typically only performed at state public health laboratories, serotype data are frequently not available to clinicians until several days after Salmonella has been identified in a clinical specimen, necessitating decisions about initial patient management in the absence of patient-specific data. Current recommendations are to treat most patients with uncomplicated Salmonella infections with supportive therapy and no antimicrobial agents [18–20]. However, understanding the relative predilection of various Salmonella serotypes for causing particular clinical syndromes has important implications for guiding public health prevention and management guidelines and further studies of the pathogenesis of these infections.

The present study did not collect data on host factors, which can have a substantial effect on the outcome of a Salmonella infection. Age of the patient, for example, could confound associations with serotype. Site of isolation of the organism was used as a proxy for invasive disease, and some persons with positive stool cultures may also have had extraintestinal disease. Data on death were not available for 18% of cases. The numbers of some serotypes available for analyses limited the ability to determine the statistical significance of observed differences in outcome. We also did not have information on the resistance patterns of isolates. With the recent increase in antimicrobial resistance in some Salmonella strains, changes in outcomes might be expected because of the failure of therapy. There is emerging evidence that antimicrobial-resistant Salmonella cause more severe infections [21–24].

This is the largest study ever published of serotype-specific outcomes of Salmonella infections, and it includes several measures of severity. It is also one of the few studies published in recent decades and provides important information pertinent to the current status of Salmonella epidemiology and therapeutic standards. Salmonella remains one of the most common enteric pathogens and causes of foodborne disease. Salmonella serotypes are closely related genetically and yet differ significantly in pathogenic potential. Understanding the mechanisms responsible for this in Salmonella may be a key to a more general understanding of the invasiveness of intestinal bacterial infections.

Potential con icts of interest: none reported.
Financial support: Centers for Disease Control and Prevention's Emerging Infections Program (cooperative agreements with participating states).

References

1.
Voetsch
AC
Van Gilder
T
Angulo
FJ
, et al.  . 
FoodNet estimate of the burden of illness caused by nontyphoidal Salmonella infections in the United States
Clin Infect Dis
 , 
2004
, vol. 
38
 
Suppl 3
(pg. 
S127
-
34
)
2.
Centers for Disease Control and Prevention
Summary of notifiable diseases—United States, 2005
MMWR
 , 
2007
, vol. 
54
 (pg. 
1
-
92
)
3.
Blaser
MJ
Feldman
RA
Salmonella bacteremia: reports to the Centers for Disease Control, 1968–1979. Centers for Disease Control
J Infect Dis
 , 
1981
, vol. 
143
 (pg. 
743
-
6
)
4.
Shimoni
Z
Pitlik
S
Leibovici
L
, et al.  . 
Nontyphoid Salmonella bacteremia: age-related differences in clinical presentation, bacteriology, and outcome
Clin Infect Dis
 , 
1999
, vol. 
28
 (pg. 
822
-
7
)
5.
Sirinavin
S
Jayanetra
P
Lolekha
S
Layangkul
T
Predictors for extra-intestinal infection in Salmonella enteritis in Thailand
Pediatr Infect Dis J
 , 
1988
, vol. 
7
 (pg. 
44
-
8
)
6.
Wittler
RR
Bass
JW
Nontyphoidal Salmonella enteric infections and bacteremia
Pediatr Infect Dis J
 , 
1989
, vol. 
8
 (pg. 
364
-
7
)
7.
Zaidi
E
Bachur
R
Harper
M
Non-typhi Salmonella bacteremia in children
Pediatr Infect Dis J
 , 
1999
, vol. 
18
 (pg. 
1073
-
7
)
8.
Cherubin
CE
Neu
HC
Imperato
PJ
Harvey
RP
Bellen
N
Septicemia with non-typhoid salmonella
Medicine (Baltimore)
 , 
1974
, vol. 
53
 (pg. 
365
-
6
)
9.
MacCready
RA
Reardon
JP
Saphra
I
Salmonellosis in Massachusetts: a sixteen-year experience
N Engl J Med
 , 
1957
, vol. 
256
 (pg. 
1121
-
8
)
10.
Cherubin
CE
Fodor
T
Denmark
LI
Master
CS
Fuerst
HT
Winter
JW
Symptoms, septicemia and death in salmonellosis
Am J Epidemiol
 , 
1969
, vol. 
90
 (pg. 
285
-
91
)
11.
Black
PH
Kunz
LJ
Swartz
MN
Salmonellosis—a review of some unusual aspects
N Engl J Med
 , 
1960
, vol. 
262
 (pg. 
811
-
7
)
12.
Allos
BM
Moore
MR
Griffin
PM
Tauxe
RV
Surveillance for sporadic foodborne disease in the 21st century: the FoodNet perspective
Clin Infect Dis
 , 
2004
, vol. 
38
 
Suppl 3
(pg. 
S115
-
20
)
13.
Saphra
I
Wassermann
M
Salmonella choleraesuis: a clinical and epidemiological evaluation of 329 infections identified between 1940 and 1954 in the New York Salmonella Center
Am J Med Sci
 , 
1954
, vol. 
228
 (pg. 
525
-
33
)
14.
Seligmann
E
Saphra
I
Wasserman
M
Salmonella infections in the USA: a second series of 2000 human infections recorded by the NY Salmonella Center
J Immunol
 , 
1946
, vol. 
54
 (pg. 
69
-
87
)
15.
Seligmann
E
Wasserman
M
Salmonella infections in man: an analysis of 1,000 cases bacteriologically identified by the New York Salmonella Center
Am J Hyg
 , 
1943
, vol. 
38
 (pg. 
226
-
49
)
16.
Saphra
I
Winter
JW
Clinical manifestations of salmonellosis in man: an evaluation of 7779 human infections identified at the New York Salmonella Center
N Engl J Med
 , 
1957
, vol. 
256
 (pg. 
1128
-
34
)
17.
Brenner
FW
Villar
RG
Angulo
FJ
Tauxe
R
Swaminathan
B
Salmonella nomenclature
J Clin Microbiol
 , 
2000
, vol. 
38
 (pg. 
2465
-
7
)
18.
Guerrant
RL
Van Gilder
TJ
Steiner
TS
, et al.  . 
Practice guidelines for the management of infectious diarrhea
Clin Infect Dis
 , 
2001
, vol. 
32
 (pg. 
331
-
50
)
19.
Thielman
NM
Guerrant
RL
Clinical practice: acute infectious diarrhea
N Engl J Med
 , 
2004
, vol. 
350
 (pg. 
38
-
47
)
20.
Centers for Disease Control and Prevention
Diagnosis and management of foodborne illnesses: a primer for physicians
MMWR
 , 
2004
, vol. 
53
 (pg. 
1
-
33
)
21.
Varma
JK
Greene
KD
Ovitt
J
Barrett
TJ
Medalla
F
Angulo
FJ
Hospitalization and antimicrobial resistance in Salmonella outbreaks, 1984–2002
Emerg Infect Dis
 , 
2005
, vol. 
11
 (pg. 
943
-
6
)
22.
Varma
JK
Molbak
K
Barrett
TJ
, et al.  . 
Antimicrobial-resistant nontyphoidal Salmonella is associated with excess bloodstream infections and hospitalizations
J Infect Dis
 , 
2005
, vol. 
191
 (pg. 
554
-
61
)
23.
Molbak
K
Human health consequences of antimicrobial drug-resistant Salmonella and other foodborne pathogens
Clin Infect Dis
 , 
2005
, vol. 
41
 (pg. 
1613
-
20
)
24.
Helms
M
Simonsen
J
Molbak
K
Quinolone resistance is associated with increased risk of invasive illness or death during infection with Salmonella serotype Typhimurium
J Infect Dis
 , 
2004
, vol. 
190
 (pg. 
1652
-
4
)