|
|
| |||||||||||||||||||||||||||||||||
Background Because Salmonella enterica serotype typhimurium is the most common serotype isolated from persons with salmonellosis in the United States, it is difficult to detect unusual clusters or outbreaks. To determine whether molecular subtyping could be useful in public health surveillance for S. enterica serotype typhimurium, the Minnesota Department of Health initiated the routine use of pulsed-field gel electrophoresis (PFGE) of isolates.
Methods Beginning in 1994, all S. enterica serotype typhimurium isolates submitted by clinical laboratories to the Department of Health were subtyped by PFGE. A standard questionnaire was used to interview patients about possible sources of infection.
Results From 1994 through 1998, 998 cases of infection with S. enterica serotype typhimurium were reported to the Minnesota Department of Health (4.4 cases per 100,000 person-years). PFGE was performed on 958 of the isolates (96 percent), and 174 different patterns were identified. Sixteen outbreaks with a common source were identified, accounting for 154 cases. PFGE subtyping made it possible to confirm 10 outbreaks that involved small numbers of cases in institutional settings. Of six larger, community-based outbreaks, four would probably not have been recognized without PFGE subtyping. These four outbreaks accounted for 96 of the 154 culture-confirmed outbreak cases (62 percent). Fifty-six of 209 isolates tested for antimicrobial susceptibility (27 percent) were resistant to at least five antimicrobial agents. The multidrug-resistant isolates identified had unique PFGE patterns.
Conclusions Routine molecular subtyping of S. enterica serotype typhimurium by PFGE can improve the detection of outbreaks and aid in the identification of multidrug-resistant strains. Combining routine molecular subtyping with a method of rapid communication among public health authorities can improve surveillance for S. enterica serotype typhimurium infections.
Methods
Ascertainment of Cases
Salmonella infections are reportable to the Minnesota Department of Health, and the public health laboratory of the department is the only laboratory that confirms and serotypes salmonella isolates from clinical laboratories within Minnesota. In October 1995, the Minnesota rules governing communicable diseases were amended to require all clinical laboratories to send salmonella isolates obtained from infected patients to the public health laboratory for the purposes of public health surveillance. In 1995, a standard questionnaire was developed and routinely administered to these patients.
A patient was defined as a Minnesota resident whose stool or blood had been found to be positive for S. enterica serotype typhimurium. Patients were identified as a result of submission of isolates from clinical laboratories or reports from health care providers. After a salmonella isolate had been confirmed as having the typhimurium serotype at the Minnesota Department of Health, the patient was interviewed by telephone with the use of the questionnaire to ascertain the history of his or her illness and potential sources of exposure to the organism during the seven days before the onset of illness (e.g., day-care attendance, drinking untreated water or unpasteurized milk, work-related exposure, and eating in a restaurant and other food-related exposure). When they were being interviewed, most patients did not know what subtype of salmonella they had been infected with.
Epidemiologic Analysis
Age-specific rates of infection were determined from 1996 population estimates obtained from Minnesota Health Statistics.10 Descriptive analyses were performed on Epi Info software (version 6.01, Centers for Disease Control and Prevention [CDC], Atlanta).
An outbreak was defined as two or more cases in different households with a common exposure. Investigations of restaurant-associated outbreaks involved interviewing employees and collecting stool specimens from them for culture. A cluster was defined as two or more cases in different households with the same subtype, occurring within two weeks of each other, for which a common source could not be determined. Outbreaks and clusters were defined as mutually exclusive.
To determine how outbreaks were detected, we divided the method of detection into two categories: subtype and interview. "Subtype" refers to investigations of outbreaks that were initiated by identification of a common cluster of subtypes on pulsed-field gel electrophoresis (PFGE). Interview-initiated investigations were identified by routine interviewing or reports from the public, health care providers, or infection-control practitioners.
To evaluate the usefulness of the Salmonella Outbreak Detection Algorithm (SODA) for the identification of S. enterica serotype typhimurium outbreaks, we compared the outbreaks that we identified with those identified by this algorithm. SODA is a laboratory-based system established in 1995 by the CDC, which compares recently reported cases of salmonella infection with a five-year mean number of cases for the same serotype and calendar week of report.11 If a statistically significant difference is found, the state health department is notified.
Molecular Subtyping
In using the PFGE method for the isolation and restriction of DNA, we followed the standard protocol of the CDC for Escherichia coli O157:H7, with minor differences in PFGE conditions.12 DNA fragments were separated with an electrophoresis apparatus (CHEF Mapper or a CHEF-DR III, Bio-Rad Laboratories, Hercules, Calif.). A strain of S. enterica serotype typhimurium (MN#E97000847) was chosen as a standard. Gels were run with the use of 0.5x TBE buffer (45 mmol TRIS base, 45 mmol boric acid, and 1 mmol EDTA) at 14°C, a linear increase in switching times (from 10.3 to 64.0 seconds) over a period of 22 hours, a 120-degree switch angle, and a gradient of 6.0 V per centimeter. The gels were then stained with 0.01 percent ethidium bromide solution (Sigma, St. Louis), digitized (Gel Doc 1000 system, Bio-Rad), and photographed with ultraviolet illumination from a fixed camera position.
Gel Analysis and Interpretation of PFGE Results
The PFGE profiles were compared by using the Jaccard coefficient and Molecular Analyst Plus software (Bio-Rad). To be considered a match, the DNA patterns could not differ from each other by more than 1 percent with respect to molecular weight. Matches were confirmed visually on the basis of the finding of exact matches of all bands in the range of 30 to 1000 kb (Figure 1). For the identification of outbreaks and clusters and for comparison of multidrug-resistant isolates, we used the guidelines of Tenover et al. to interpret DNA-restriction patterns generated by PFGE.13
|
Beginning in 1995, a program monitoring antimicrobial resistance was established. Every fifth salmonella isolate received by the public health laboratory was tested for antimicrobial-drug susceptibility. Isolates from nonresidents of Minnesota, duplicate isolates, and all but the first isolate from investigations of outbreaks were removed from the analysis. Disk-diffusion tests were performed on Mueller–Hinton agar with standard zone diameters established by the National Committee for Clinical Laboratory Standards.14 The antimicrobial-susceptibility panel included ampicillin, cefotaxime, cephalothin, chloramphenicol, ciprofloxacin, gentamicin, kanamycin, nalidixic acid, streptomycin, tetracycline, and trimethoprim–sulfamethoxazole. Sulfisoxazole was added to the panel in 1998, and the susceptibility of selected multidrug-resistant isolates was evaluated. Selected multidrug-resistant isolates were sent to the CDC for bacteriophage typing.
Results
Characteristics of the Isolates
From 1994 through 1998, 998 cases of S. enterica serotype typhimurium were reported among Minnesota residents (4.4 cases per 100,000 person-years). The highest age-specific rate was among children younger than five years of age: 13.5 cases per 100,000 person-years. A total of 450 of the 998 cases (45 percent) were identified from specimens collected during the four months from June through September in any given year. From 1995 through 1998, 771 of 864 patients (89 percent) were interviewed. The remaining patients were lost to follow-up because they could not be contacted, they declined to be interviewed, or there were language barriers.
Isolates were received from 958 reported cases (96 percent) from 1994 through 1998 (Table 1). One hundred seventy-four PFGE patterns were identified, of which 106 (61 percent) were represented by only a single case isolate. The five most common patterns accounted for 448 isolates (47 percent). These five patterns were identified each year during the five-year period.
|
Detection of Outbreaks
No outbreaks were detected in 1994. From 1995 through 1998, a total of 16 common-source outbreaks and 63 clusters were identified (Table 2). These accounted for 154 and 241 cases, respectively. Of the 16 common-source outbreaks, 10 occurred in institutional settings and were detected by either telephone interviews of patients by staff members of the Minnesota Department of Health or telephoned reports from the public, physicians, or infection-control practitioners (Table 3). These included six outbreaks among children in a home child-care setting; two among nursing-home residents; one among students, faculty, and horses at a veterinary teaching hospital; and one involving two patients in a renal dialysis unit. The 10 outbreaks involved small numbers of cases, which PFGE subtyping confirmed as outbreak-associated rather than unrelated sporadic cases. This prompted a review of infection-control practices in each setting.
|
|
Six outbreaks were in community settings. Four involved restaurants, one involved contaminated pork served at a wedding reception, and the other involved a commercial microwavable chicken product sold in grocery stores.15 These food-borne outbreaks accounted for 128 of the 154 outbreak-related cases (83 percent). PFGE subtyping was instrumental in initiating four of these six outbreak investigations, which accounted for 96 of the 154 culture-confirmed outbreak cases (62 percent). For example, over a 15-week period in 1998, 79 cases of S. enterica serotype typhimurium infection were reported to the Minnesota Department of Health. Initial interviews of patients identified only two outbreaks, each consisting of two cases, one outbreak associated with a wedding reception and the other with a child-care setting. However, an indistinguishable PFGE pattern was observed among 32 of the 79 isolates (41 percent). More focused interviews were performed with patients with this PFGE pattern, and an outbreak associated with a commercial microwavable chicken product was identified, leading to the recall of the product and changes in the label to identify the product more clearly as uncooked.15
PFGE subtyping helped us to set priorities and focused efforts when increases in multiple patterns were concurrently observed, especially during the peak summer months. From 1995 through 1998, 10 outbreaks overlapped temporally. In eight outbreaks, cases were identified over a period of at least four weeks, with a median of two cases identified per week (range, zero to seven). In 1995, three restaurant-associated outbreaks occurred simultaneously (Figure 2). Two of these outbreaks were identified through subtype-specific surveillance, and the other was reported to the Minnesota Department of Health by a hospital infection-control practitioner. During a 16-week period, 157 cases of S. enterica serotype typhimurium infection were reported. The outbreak-associated PFGE patterns from these restaurants accounted for 55, 20, and 24 cases. In initial interviews of the patients infected during the two restaurant outbreaks identified through subtype-specific surveillance, fewer than half of the patients reported eating at the restaurant ultimately associated with infection.
|
From 1996 through 1998, the SODA system of the CDC identified nine instances in which there was a statistically significant increase in the number of cases of S. enterica serotype typhimurium infection occurring in a given week and notified the Minnesota Department of Health. In six of the instances, there was no evidence of an outbreak; rather, increases were due to an increase in the number of isolates with multiple, unrelated PFGE patterns. In one instance, two outbreaks occurred simultaneously. The remaining two notifications were for a single extended outbreak involving a commercial food product.15 From 1996 through 1998, the Minnesota Department of Health detected 11 outbreaks of infection due to S. enterica serotype typhimurium. The SODA system notified the Minnesota Department of Health of three of these outbreaks two to four weeks after the Minnesota Department of Health had already identified them.
Antimicrobial-Susceptibility Profiles
Two hundred nine isolates from Minnesota residents were tested for antimicrobial susceptibility. All isolates were sensitive to cefotaxime and ciprofloxacin. Fifty-six of the 209 isolates (27 percent) were resistant to five or more antimicrobial agents. Twenty-eight (13 percent) were resistant to at least ampicillin, chloramphenicol, streptomycin, sulfisoxazole, and tetracycline (R-type ACSSuT). These were represented by 13 different PFGE patterns, of which 11 (85 percent) were clonally related. The 11 PFGE patterns accounted for 26 of the 28 isolates (93 percent). Two PFGE patterns accounted for 17 of the 28 isolates (61 percent) (Figure 1). One of these two patterns was consistently observed over the five-year period (lane 4 in Figure 1). Nineteen isolates that were resistant to ampicillin, chloramphenicol, streptomycin, sulfisoxazole, and tetracycline underwent phage typing, and 18 (95 percent) were part of the definitive type 104 (DT104) complex.
Of the remaining 28 isolates resistant to five or more antimicrobial agents, 21 (75 percent) were resistant to ampicillin, kanamycin, streptomycin, sulfisoxazole, and tetracycline (R-type AKSSuT). Fifteen of the 21 isolates (71 percent) were represented by a single PFGE pattern, which was observed each year (lane 6 in Figure 1). This single pattern accounted for 16 isolates during this time, of which 94 percent were resistant to ampicillin, kanamycin, streptomycin, sulfisoxazole, and tetracycline. The results of phage typing of selected isolates with this pattern did not conform to any known phage types.
Discussion
Routine molecular subtyping of isolates of S. enterica serotype typhimurium greatly enhanced our surveillance for this common serotype of salmonella. With knowledge of the PFGE pattern, we were able to assign priorities for treating patients with S. enterica serotype typhimurium infection, focus investigations, and avoid unnecessary investigations of concurrent increases in unrelated patterns. Routine subtyping allowed us to detect outbreaks that would not have been detected by traditional surveillance methods, as well as to confirm outbreaks reported by traditional methods, since we could identify cases as outbreak-associated rather than as unrelated sporadic cases. This approach allowed us to intervene to stop transmission from a variety of sources of infection and to rapidly identify specific multidrug-resistant strains of importance to public health.
We identified 174 unique PFGE patterns of S. enterica serotype typhimurium over a five-year period. The diversity of PFGE subtypes probably indicates that infections were derived from many different sources. These PFGE patterns were stable over time. There were common subtype patterns over the entire five-year period, with homogeneity of serial stool cultures from individual patients and among outbreak-related strains. PFGE patterns were also associated with antimicrobial-resistance profiles and phage-type results.
The routine use of PFGE and the epidemiologic follow-up of individual patients were crucial to identify and confirm the 16 outbreaks detected between 1995 and 1998. Four of six community-based outbreaks would not have been detected without subtype-specific surveillance. In one instance, subtype-specific surveillance was instrumental in the recall of a contaminated commercial product. Before the introduction of PFGE surveillance, 563 cases of S. enterica serotype typhimurium infection were reported to the Minnesota Department of Health from 1990 through 1993, but no outbreaks of this serotype were recognized. In half the outbreaks in our study, we identified a median of two cases per week. The onset of illness during these outbreaks occurred over a period of weeks or months. We believe these cases resulted from low-level, intermittent transmission of S. enterica serotype typhimurium through contact with contaminated food, infected food workers, and environmental contamination in restaurants and licensed home child-care settings.
Conventional surveillance systems look for unusual temporal increases in serotypes and are likely to miss many outbreaks of common serotypes, such as S. enterica serotype typhimurium, S. enterica serovar enteritidis, and S. enterica serotype heidelberg. Conversely, conventional surveillance leads to unnecessary investigations of unrelated cases that are temporally clustered. These systems lack the sensitivity to detect outbreaks involving common serotypes, especially when they involve few cases occurring over a period of several weeks. This is also true for automated algorithms, such as SODA. Twenty-seven percent of clusters identified by the SODA system in 1997 that prompted notification to states were serotype typhimurium.16 For surveillance of S. enterica serotype typhimurium infections in Minnesota, this algorithm was neither sensitive nor specific in detecting outbreaks. One way to increase specificity of this algorithm would be to include information on molecular subtypes for common serotypes in each state. Because the usefulness of the SODA system is dependent on the timeliness of data entry, states should be encouraged to update this information daily.
Surveillance systems for S. enterica serotype typhimurium should include routine molecular subtyping together with the epidemiologic support to investigate subtype clusters. The routine use of PFGE subtyping requires isolates to be sent to the state public health laboratory promptly and individual patients to be interviewed. Isolates should be typed as soon as possible after receipt rather than in batches, and retrospective summaries detailing PFGE patterns according to the week, month, and year should be provided. In Minnesota, isolates were received at the Minnesota Department of Health a median of five days after stool collection, and the PFGE results were generally available two to seven days later.
Surveillance of molecular subtypes was also useful in identifying specific multidrug-resistant strains of S. enterica serotype typhimurium. Previous investigators have highlighted the use of plasmid analysis, PFGE, and phage typing for differentiating multidrug-resistant S. enterica serotype typhimurium.17,18,19,20 We found that isolates with certain unique PFGE patterns consistently had common phage types (e.g., DT104) and antimicrobial-resistance profiles (e.g., they were resistant to ampicillin, chloramphenicol, streptomycin, sulfisoxazole, and tetracycline [R-type ACSSuT] or resistant to ampicillin, kanamycin, streptomycin, sulfisoxazole, and tetracycline [R-type AKSSuT]). These characteristics have also been observed by Besser et al.,21 and we recently found this association in an outbreak of S. enterica serotype typhimurium DT104 in a home child-care setting.22 Phage typing has been the primary method of subtyping for S. enterica serotype typhimurium, but it requires access to special reagents and can only be performed at a few institutions. Many state public health laboratories can perform PFGE, and in the absence of timely phage typing, PFGE is a simple way to detect and monitor multidrug-resistant strains.
We have previously demonstrated that PFGE is a crucial tool for public health surveillance and disease-prevention efforts for E. coli O157:H7.23 Our results support the same conclusion with respect to surveillance for S. enterica serotype typhimurium and most likely other common salmonella serotypes. Recently, the CDC suggested that all state public health laboratories should have the capacity to perform molecular subtyping of food-borne pathogens in a timely manner. PulseNet, the CDC's national molecular-subtyping network for food-borne disease surveillance, currently uses PFGE to characterize E. coli O157, Listeria monocytogenes, shigella, and salmonella.
Continued efforts are needed to increase the number of states in which routine rapid molecular subtyping is performed. Routine subtyping and epidemiologic follow-up will also demonstrate to clinical laboratories the value of submitting isolates to public health laboratories. Without the participation of clinical laboratories, on either a voluntary or a mandatory basis, public health surveillance for food-borne pathogens will be inadequate. As more states are included in the PulseNet system, detection of regional outbreaks of infection with S. enterica serotype typhimurium and identification of contaminated food products will increase.
Supported in part by a cooperative agreement (U50/CCU511190) with the Centers for Disease Control and Prevention.
Source Information
From the Acute Disease Epidemiology Section (J.B.B., C.W.H., J.H.W., K.E.S., M.T.O.) and the Division of Public Health Laboratories (D.J.B., J.M.B.), Minnesota Department of Health, Minneapolis.
Address reprint requests to Dr. Bender at the University of Minnesota, 1988 Fitch Ave., St. Paul, MN 55108, or at bende002{at}tc.umn.edu.
References
| |||||||||||||||||||||||||||||||||
This article has been cited by other articles:
HOME | SUBSCRIBE | SEARCH | CURRENT ISSUE | PAST ISSUES | COLLECTIONS | PRIVACY | TERMS OF USE | HELP | beta.nejm.org Comments and questions? Please contact us. The New England Journal of Medicine is owned, published, and copyrighted © 2009 Massachusetts Medical Society. All rights reserved. |
Previous
