Molecular epidemiologic surveillance of salmonellosis in Arkansas.

Objectives: Salmonella serotype Newport and Salmonella serotype Typhimurium are the most commonly identified serotypes of Salmonella causing human disease in the state of Arkansas. The purpose of our study was to compare the results of standard and molecular epidemiologic methods of investigating human salmonellosis cases due to Salmonella serotype Newport and Salmonella serotype Typhimurium.

Methods: All isolates of Salmonella serotype Newport and Salmonella serotype Typhimurium collected and submitted to the Arkansas Department of Health between July 1, 1997 and June 30, 1998 were gathered and underwent pulsed-field gel electrophoresis (PFGE). Patients from whom the isolates were collected were contacted and completed a questionnaire.

Results: There were 84 patients from whom Salmonella serotype Newport was isolated and 83 from whom Salmonella serotype Typhimurium was isolated during the study period. In the 124 patients (74%) who completed the questionnaire. Salmonella serotype Newport was more likely to be the infecting agent in younger, white, and pet-owning patients (P < 0.05). The use of PFGE confirmed that approximately 20% of the organisms had genetic fingerprint patterns identical to those of at least one other individual in the state. One third of the patients from whom these isolates were obtained were linked by standard epidemiologic methods.

Conclusions: The use of PFGE on our state's most common isolates provides additional confirmation that despite being linked by time of onset and location of residence, the majority of the human salmonellosis cases in our region are still sporadic. Low-level, intermittent transmission of these organisms through environmental contamination and contact with asymptomatically infected individuals would be likely vehicles of transmission in our state. Molecular techniques are important in surveillance systems that investigate human salmonellosis. Eighty-one percent of the Salmonella serotype Newport and 92% of the Salmonella serotype Typhimurium cases that appeared to be outbreak-related based upon time of onset and location were actually found not to be outbreak-related by PFGE. Using techniques such as PFGE will allow for more focused evaluations of potential outbreaks and will save the already limited financial and human resources that would otherwise be spent on investigations that are not warranted.

Key Words: epidemiology, pulsed-field gel electrophoresis, Salmonella

**********

Salmonella serotype Newport and Salmonella serotype Typhimurium are the two most commonly identified serotypes of Salmonella causing human disease in the state of Arkansas. (1) The majority of these infections are thought to be sporadic cases of human disease. However, acute outbreaks with either of these two serotypes are difficult to identify using standard epidemiologic tools unless a sharp increase in cases from a localized region is noted. The ability to have an accurate method with which to analyze these serotypes could be extremely important to the health of the inhabitants of the state.

Recently, it has been demonstrated that pulsed-field gel electrophoresis (PFGE) is useful in identifying genetically identical nontyphoidal Salmonella isolates including Salmonella serotype Newport and Salmonella serotype Typhimurium specifically. (2-10) This technique can be useful in distinguishing outbreak-related from non-outbreak-related isolates, and can be helpful in guiding a more focused epidemiologic investigation. (5,9) The purpose of the current study was to apply both standard and molecular epidemiologic methods to investigating human salmonellosis cases, and to compare the results obtained by using the two methods.

Materials and Methods

All isolates of Salmonella serotype Newport and Salmonella serotype Typhimurium submitted to the Arkansas Department of Health were gathered and underwent DNA fingerprinting by PFGE using previously described techniques. (7) The isolates were collected from patients from July 1, 1997, throug June 30, 1998. Restriction enzyme digestion was carried out using both XbaI (5'-TCTAGA-3') and AvrII (5'-CCTAGG-3') (Fig. 1). Fingerprints of like isolates were compared by using Bio-Rad Molecular Analyst/PC fingerprinting software (Bio-Rad Laboratories, Hercules, CA). Because the isolates being compared were of the same serotype, their PFGE patterns were expected to be less polymorphic than those of other species of Enterobacteriaceae as defined by biotyping or ribosomal gene sequencing. Consequently, they were considered to be genotypically different if the PFGE pattern differed by a single fragment. This definition is more restrictive than that applied to other organisms collected from a restricted locale over a short period. (11) It is justified based on the fact that the organisms were collected from an area enclosing over 50,000 square miles over a 1-year period, and the organisms compared are limited to one of approximately 2,000 serotypes of the species Salmonella. All DNA fingerprint patterns were recorded and stored. Isolates with similar PFGE patterns were restricted and electrophoresed on the same gel to confirm identity. To be considered a match, the isolates from two or more patients required a complete agreement of all bands in the range of 30 to 1,000 kilobase (kb) by PFGE after digestion with both XbaI and AvrII. All other isolates were considered to have unique PFGE patterns.

[FIGURE 1 OMITTED]

Attempts were made to telephone all patients with salmonellosis. Once contact was established and after informed consent was obtained, basic demographic data were verified and a questionnaire was administered by a single trained interviewer (E.L.F.). Questions included but were not limited to employment, food exposure and consumption, travel, water source, and sick contacts. Data were reviewed in an attempt to connect cases by time of onset and place of residence or travel. Statistical comparisons were made by using the Wilcoxon rank-sum test for continuous data and the Fisher exact test or Pearson [chi square] test for categorical data.

Results

During the study period, 191 isolates of Salmonella serotype Newport and Salmonella serotype Typhimurium were sent to the Arkansas Department of Health. This represented 54% of all Salmonella isolates that were reported in the state (Salmonella serotype Newport, 28%; Salmonella serotype Typhimurium, 26%). Twenty-four isolates were excluded from further evaluation because 20 (Salmonella serotype Newport, 11; Salmonella serotype Typhimurium, 9) were repeat positive cultures from individuals in the current study and 4 were from environmental sources (eg, dirt) from a previous study (Salmonella serotype Newport, 3; Salmonella serotype Typhimurium, 1). (8) The remaining 167 isolates were obtained from separate individuals (Salmonella serotype Newport, 84; Salmonella serotype Typhimurium, 83). An attempt was made to complete the questionnaire in all 167 patients but was unsuccessful in 43 (26%) for the following reasons: the phone had been disconnected (n = 18), contact could not be made with the family after four phone calls (n = 15), or the individuals refused to participate (n = 8) or could not communicate through the telephone (n = 2). Although basic demographic data and the DNA fingerprinting of the patient isolates were available for comparison to other isolates in these 43 patients, further epidemiologic information was based on the 124 subjects that completed the questionnaire (Salmonella serotype Newport, 65; Salmonella serotype Typhimurium, 59).

The average age of patients was 11.1 [+ or -] 18.3 years ([+ or -] SD) for those infected with Salmonella serotype Newport and 17.6 [+ or -] 24.7 years for those infected with Salmonella serotype Typhimurium. Demographic data for patients infected with Salmonella serotype Newport or Salmonella serotype Typhimurium are summarized in Table 1. In general, patients infected with Salmonella serotype Newport were younger (P < 0.05), more likely to be white (P < 0.01), and have a history of pet ownership (P < 0.02) as compared with patients infected with Salmonella serotype Typhimurium. There were no consistent types of pets that were owned except that the majority were mammals.

Reviewing the epidemiologic data gathered using time of onset and location of residence, six groups of Salmonella serotype Newport involving 37 individuals and eight groups of Salmonella serotype Typhimurium involving 26 individuals were identified. Within each group, all human infections were found to have occurred within 1 month of each other and could be related by city of residence or travel. Using these traditional methods, 57% of the Salmonella serotype Newport and 44% of Salmonella serotype Typhimurium infections appeared to be from a single-exposure, common-vehicle outbreak.

For Salmonella serotype Newport, there were 72 different PFGE patterns identified. DNA fingerprinting was not successful in one isolate. There were 18 patients that could be grouped with seven distinct matched patterns. Three of these groups involved seven patients that were previously identified by time of onset and location. Two additional patients were added to one of these groups and four additional groups involving nine patients were identified only by matched PFGE patterns. For Salmonella serotype Typhimurium, 74 different PFGE patterns were identified, whereas there were 15 patients that could be grouped with six distinct matched patterns. Only two of these patients in one group were identified by time of onset and location, whereas matched PFGE patterns added two additional patients to this group and identified five other groups. The remaining patients infected with either serotype that appeared to have a common source on the basis of time of onset and locations all had unique DNA fingerprints and were genotypically different.

For the patients with Salmonella serotype Newport that had matched patterns by PFGE, all were white (100%) and 16 (89%) were 13 years of age or younger. The remaining two patients were 31 and 59 years of age. Daycare was identified as the common risk factor for infection in two groups (seven patients), whereas in the third group (two patients), the patients were brother and sister. In three of the remaining groups (seven patients), cultures were obtained from the patients during the same week, even though there was a varying amount of distance that separated them (19-221 miles). In the remaining group (two patients), approximately 10 weeks separated the culture dates and, although no connection was noted, these patients lived within 20 miles of each other.

Of the patients with Salmonella serotype Typhimurium with matched patterns by PFGE, 10 were white and 10 (67%) were 13 years of age or younger. The largest group involved four patients. Two of these were husband and wife and a third individual lived in the same county but in a town approximately 29 miles away. The isolate from the fourth patient was recovered approximately 1 month later from a 3-year-old, 260 miles from the others. Only one of the five remaining groups (two patients) occurred during the same week. Neither of these patients completed the questionnaire, so no further data were available. The remaining four groups (nine patients) occurred from 11 days to 7 months apart and were located 70 to 125 miles apart. In none of the groups of matched patterns of Salmonella serotype Newport or Salmonella serotype Typhimurium could the initial source of the infection be identified.

Discussion

The use of PFGE on our state's most common isolates provides additional confirmation that the majority of the human salmonellosis cases in our region are still sporadic. Using traditional surveillance methods, 57% of Salmonella serotype Newport and 44% of Salmonella serotype Typhimurium infections that occurred in our state appeared to be from a common source on the basis of time of onset of disease and location of residence or travel. The use of DNA fingerprinting by PFGE proved that 81% of the Salmonella serotype Newport and 92% of the Salmonella serotype Typhimurium cases did not have a match that was previously related using traditional methods. Although the use of PFGE has been quite successful in connecting nontyphoidal Salmonella isolates in large outbreaks, (2-5,10) there are limited data attempting to evaluate an area with PFGE on endemic isolates. Previous data collected in our region on salmonellosis cases in children younger than 4 years of age demonstrated that 35% of the 29 isolates of Salmonella serotype Newport and 9% of the 22 isolates of Salmonella serotype Typhimurium could be linked by PFGE. (7) Using the larger number of isolates in the current study, 21% of the Salmonella serotype Newport and 18% of the Salmonella serotype Typhimurium isolates could be matched to at least one other person.

In this study, the criterion for PFGE identity was more stringent than that routinely applied to other organisms. To be considered a match, the isolates from two or more patients required complete agreement of all bands by PFGE after digestion with two restriction endonucleases. In more limited studies on other organisms, the term "closely related" is used to characterize patterns that differ by one to three restriction fragments. (11) Such organisms (eg, Staphylococcus aureus, Escherichia coli, Enterococcus species) are not routinely subspeciated, as are the Salmonella, by serotyping. Moreover, Tenover et al (11) have stated that "the criteria for strain identity are stringent and are not appropriate for studies of large populations of organisms collected over extended periods of one year or longer." For these reasons, the criterion for strain identity was modified to include only those that were an identical match with two enzymes rather than including those that differed by one to three fragments with one restriction enzyme.

For patients with matched isolates, the majority could be related by time of onset and/or location, which would suggest a common source. Those patients infected with Salmonella serotype Newport were easier to relate to each other than those infected with Salmonella serotype Typhimurium. Unless the patients had an apparent risk factor (eg, daycare, same family), the use of traditional surveillance methods failed to identify a common source of infection for either serovar. In a recent study involving 958 isolates of Salmonella serotype Typhimurium obtained in Minnesota over a 4-year period, 395 isolates (41%) had matches. (5) There were 154 cases (39%) accounted for by a common risk factor (eg, nursing home, daycare), whereas 241 cases (61%) involving 63 clusters of matched isolates did not have a common risk factor identified. The data presented here could identify matches in 18% of Salmonella serotype Typhimurium isolates, with 13% having a common risk factor and 87% not having a risk factor or source identified. For Salmonella serotype Newport, 21% of isolates had matches, and 50% had a common risk factor identified and 50% did not.

The differences noted here between racial makeup and history of pet ownership across the two serotypes have not been noted in our previous studies, and their meaning is unclear. The propensity for Salmonella serotype Newport to infect younger patients has been noted in previous work without a good explanation. (7,8) For patients who did not have a readily recognized risk factor of infection (eg, daycare), no common factors (eg, food) could be identified for their infection despite being infected with a matched serotype of Salmonella.

Low-level, intermittent transmission of these organisms through environmental contamination (1,7,8,12-15) or contact with asymptomatically infected individuals (1,7,8,16-21) would be likely vehicles of transmission in our state. Contaminated foods certainly play a significant role in large outbreaks of human salmonellosis, (4,22) but foods have never been proven to be the cause of spreading sporadic human infections in our region. (7,8) Recent data have suggested that contaminated eggs remain important vehicles for the transmission of Salmonella, even in the absence of recognized outbreaks. (23) However, our previous investigations have demonstrated that contaminated poultry eggs are not a major risk factor in our state. (24)

These data underscore the importance of molecular techniques in surveillance systems in the investigation of salmonellosis. By using molecular subtyping on commonly encountered isolates in addition to traditional epidemiology, state health officials will be able to detect outbreaks that would not have been identified otherwise and will allow confirmation of those identified through the traditional methods. In addition, this will allow for more focused evaluations of potential outbreaks and will save the limited financial and human resources on investigations that are not warranted.

Table 1. Demographic data and risk factors of 124 patients infected with
Salmonella serotype Newport (n = 65) and Salmonella serotype Typhimurium
(n = 59)

                                     Salmonella  Salmonella
                                      serotype    serotype
                                       Newport   Typhimurium
Variable                                 (%)        (%)

Race
  White                                  90          66 (a)
  Black                                   9          27
Gender
  Male                                   46          44
  Female                                 54          56
Water source
  Well water                             11           9
  City water                             81          84
  Bottled water                           8           7
Risk factors
  Others sick outside the home           14          14
  Others sick inside the home            15          15
  Pet ownership                          74          53 (b)
  Livestock ownership                     9          14
  Travel outside of hometown             30          29
  Food prepared outside of the home      59          66

(a) P < 0.01.
(b) P < 0.02.

Accepted October 22, 2003.

Copyright [c] 2004 by The Southern Medical Association

0038-4348/04/9706-0583

References

1. Schutze GE, Flick EL, Pope SK, et al. Epidemiology of salmonellosis in Arkansas. South Med J 1995;88:195-199.

2. L'Ecuyer PB, Diego J, Murphy D, et al. Nosocomial outbreak of gastroenteritis due to Salmonella senftenberg. Clin Infect Dis 1996;23:734-742.

3. Puohiniemi R, Heiskanen T, Siitonen A. Molecular epidemiology of two international sprout-borne Salmonella outbreaks. J Clin Microbiol 1997;35:2487-2491.

4. Taylor JP, Barnett BJ, del Rosario L, et al. Prospective investigation of cryptic outbreaks of Salmonella agona salmonellosis. J Clin Microbiol 1998;36:2861-2864.

5. Bender JB, Hedberg CW, Boxrud DJ, et al. Use of molecular subtyping in surveillance for Salmonella enterica serotype typhimurium. N Engl J Med 2001;344:189-195.

6. Murase T, Okitsu T, Suzuki R, et al. Evaluation of DNA fingerprinting by PFGE as an epidemiologic tool for Salmonella infections. Microbiol Immunol 1995;39:673-676.

7. Schutze GE, Kirby RS, Flick EL, et al. Epidemiology and molecular identification of Salmonella infections in children. Arch Pediatr Adolesc Med 1998;152:659-664.

8. Schutze GE, Sikes JD, Stefanova R, et al. The home environment and salmonellosis in children. Pediatrics 1999;103:el.

9. Giammanco GM, Pignato S, Mammina C, et al. Persistent endemicity of Salmonella bongori 48:[z.sub.35]:--in Southern Italy: Molecular characterization of human, animal, and environmental isolates. J Clin Microbiol 2002;40:3502-3505.

10. Winthrop KL, Palumbo MS, Farrar JA, et al. Alfalfa sprouts and Salmonella kottbus infection: A multistate outbreak following inadequate seed disinfection with heat and chlorine. J Food Prot 2003; 66:13-17.

11. Tenover FC, Arbeit RD, Goering RV, et al. Interpreting chromosomal DNA restriction patterns produced by pulsed-field gel electrophoresis: Criteria for bacterial strain typing. J Clin Microbiol 1995;33:2233-2239.

12. Angulo FJ, Tippen S, Sharp DJ, et al. A community waterborne outbreak of salmonellosis and the effectiveness of a boil water order. Am J Public Health 1997;87:580-584.

13. Burton GA Jr, Gunnison D, Lanza GR. Survival of pathogenic bacteria in various freshwater sediments. Appl Environ Microbiol 1987;53:633-638.

14. Haddock RL. Salmonella in vacuum cleaners [letter]. Lancet 1986;2:637.

15. Haddock RL, Malilay J. A search for infant salmonellosis risk factors on Guam. Southeast Asian J Trop Med Public Health 1986;17:38-42.

16. Buchwald DS, Blaser MJ, A review of human salmonellosis: Part II. Duration of excretion following infection with nontyphi Salmonella. Rev Infect Dis 1984;6:345-356.

17. Deutch M, Sonnenwirth A. Salmonellosis in private practice: How helpful are stool cultures? Clin Pediatr (Phila) 1965;35:511-514.

18. Rosenstein BJ. Salmonellosis in infants and children. J Pediatr 1967;70:1-7.

19. Szanton VL. Epidemic salmonellosis: A 30-month study of 80 cases of Salmonella oranienburg infection. Pediatrics 1957;20:794-808.

20. van Schothorst M, Huisman J, van Os M. Search for salmonellas in homes with salmonellosis in infants [in Dutch]. Ned Tijdschr Geneeskd 1978;122:1121-1125.

21. Wilson R, Feldman RA, Davis J, et al. Salmonellosis in infants: The importance of intrafamilial transmission. Pediatrics 1982;69:436-438.

22. Wegener HC, Baggesen DL. Investigation of an outbreak of human salmonellosis caused by Salmonella enterica ssp. enterica serovar Infantis by use of pulsed field gel electrophoresis. Int J Food Microbiol 1996;32:125-131.

23. Hedberg CW, David MJ, White KE, et al. Role of egg consumption in sporadic Salmonella enteritidis and Salmonella typhimurium infections in Minnesota. J Infect Dis 1993;167:107-111.

24. Schutze GE, Fawcett HA, Lewno MJ, et al. Prevalence of Salmonella enteritidis in poultry shell eggs in Arkansas. South Med J 1996;89:889-891.

RELATED ARTICLE: Key Points

* Molecular techniques will greatly advance the study of outbreaks of Salmonella infections.

* Pulsed-field gel electrophoresis is a molecular technique that is beneficial for evaluating outbreaks of Salmonella.

Gordon E. Schutze, MD, Ellie L. Flick, MSED, Rossina Stefanova, PHD, H.J. Spencer, MS, Shelly Y. Lensing, MS, Dennis A. Berry, MS, and M. Donald Cave, PHD

From the Departments of Pediatrics, Pathology, and Anatomy, School of Medicine, University of Arkansas for Medical Sciences; and the Arkansas Department of Health, Little Rock, AR.

This work was sponsored through funding from the Food Safety Consortium that is supported by the United States Department of Agriculture. The University of Arkansas, Agricultural Experimental Station, Fayetteville, AR, issued this grant.

This study was reviewed and approved by the Human Research Advisory Board of the University of Arkansas for Medical Sciences in Little Rock, AR.

Reprint requests to Gordon E. Schutze, MD, Arkansas Children's Hospital, 800 Marshall Street, Little Rock, AR 72202-3591. Email: [e-mail]