|
|
| |||||||||||||||||||||||||||||||||
Background Escherichia coli O157:H7 is a leading cause of diarrhea and the hemolytic–uremic syndrome. Current public health surveillance for E. coli O157:H7 requires considerable resources; traditional methods lack the sensitivity and specificity to detect outbreaks effectively.
Methods During 1994 and 1995, the Minnesota Department of Health requested that all clinical isolates of E. coli O157:H7 be submitted to our laboratory. Isolates were subtyped by pulsed-field gel electrophoresis (PFGE), and patients were interviewed about potential sources of infection.
Results In 1994 and 1995, 344 cases of E. coli O157:H7 infection were reported to the Minnesota Department of Health; 317 (92 percent) were subtyped by PFGE, and 143 distinct PFGE patterns were identified. Ten outbreaks of E. coli O157:H7 were identified; these accounted for 56 (18 percent) of the 317 subtyped cases. Four outbreaks were detected solely as a result of subtype-specific surveillance. In 11 two-week periods, the number of reported cases of E. coli O157:H7 doubled from the previous two weeks. In eight of these instances, the patterns identified were dissimilar and there were no outbreaks. Two of the remaining three increases resulted from multiple simultaneous outbreaks.
Conclusions Subtype-specific surveillance for E. coli O157:H7 can identify outbreaks that are not detected by traditional methods and can ascertain whether sudden increases in reported cases are due to sporadic isolated cases or to one or more outbreaks.
Molecular-subtyping methods have been developed to discriminate various strains of E. coli O157:H7.3,4,5,6,7 These methods have been applied primarily when there have been outbreaks, to distinguish between outbreak-associated and sporadic or unrelated infections. When there was a sudden increase in reported cases, subtyping was used to discriminate among isolates of E. coli O157:H7, helping the epidemiologic investigators to identify a common vehicle of infection.8 However, the value of isolate subtyping as a surveillance tool on a population basis is not known. We began routine molecular subtyping of E. coli O157:H7 isolates submitted to the Minnesota Department of Health in 1994 and 1995. The objectives of our study were to improve our understanding of the epidemiology of E. coli O157:H7 and to determine whether routine molecular subtyping of E. coli O157:H7 isolates is useful in public health practice.
Methods
Case Ascertainment
The reporting of E. coli O157:H7 infections to the Minnesota Department of Health became compulsory in 1987. In 1994 and the first three quarters of 1995, the department requested that all clinical laboratories in the state submit any E. coli O157:H7 isolates from Minnesota residents to the state public health laboratory for confirmation and molecular subtyping. Beginning in October 1995, submission of E. coli O157:H7 isolates was required under the Minnesota rules governing communicable diseases.
Patients were defined as sick people who resided in Minnesota and had E. coli O157:H7 isolated from a stool culture. They were identified through the submission of isolates from clinical laboratories or through direct reports from health care providers. Once isolates were confirmed as being E. coli O157:H7, the patients were interviewed by telephone; a standardized questionnaire was used to ascertain their history of diarrheal illness and any potential common exposures (e.g., attending a day-care center, drinking untreated water, consuming undercooked hamburger, and eating in restaurants). Interviews were conducted before the results of subtyping by pulsed-field gel electrophoresis (PFGE) became available, by interviewers who were not aware of the results.
The reporting of the hemolytic–uremic syndrome was required in Minnesota beginning in 1987. Patients were defined as Minnesota residents with the triad of microangiopathic hemolytic anemia (hematocrit <30 percent, with evidence of intravascular erythrocyte destruction on a peripheral-blood smear), thrombocytopenia (platelet count, <150,000 per cubic millimeter), and evidence of renal involvement (blood urea nitrogen, >20 mg per deciliter or abnormal urinary sediment). Patients were identified through an active–passive reporting system. The active component was based on routine contact with specialists in pediatric intensive care and infection-control practitioners at major tertiary care centers in the state. In addition, Minnesota physicians were encouraged to obtain stool cultures from all patients with the hemolytic–uremic syndrome for isolation of E. coli O157:H7.
Epidemiologic Analysis
Information collected for each patient included his or her name, age, sex, and place of residence; the molecular-subtype designation of the isolate; the "event date"; the presence or absence of the hemolytic–uremic syndrome; and whether the case was part of a recognized outbreak. The event date was defined as the earliest date for which recorded information was available about the onset of symptoms, the date of stool collection, or the date the specimen was received at the laboratory. To determine overall and age-specific rates of infection, population estimates were obtained from Minnesota Health Statistics, 1994.9 Descriptive analyses were conducted by using EpiInfo, version 6.01 (CDC, Atlanta).
An outbreak was defined as two or more cases in separate households with a common epidemiologic exposure. A case cluster was defined as two or more cases involving the same subtype, occurring within two weeks of each other, for which no common source could be determined by epidemiologic investigation. Temporal increases in the number of cases were evaluated by comparing the number of cases detected during the current two-week period with the number of cases detected during the previous two-week period.
Molecular Subtyping
DNA Isolation and Restriction
Bacterial suspensions were made by emulsifying the growth from 24-hour sheep's-blood agar cultures (BHI, Becton Dickinson, Cockeysville, Md.) in physiologic saline and standardizing to 55 to 65 percent transmittance at 590 nm. The suspension was concentrated by centrifugation, resuspended in EET (100 mM EDTA, 10 mM ethylene glycol-bis[beta-aminoethyl ether]N,N,N',N'-tetraacetic acid, and 10 mM Tris; pH 8.0), and immobilized in 1.6 percent agarose (SeaPlaque, GTG; FMC Bioproducts, Rockland, Me.). Plugs were then treated with lysozyme (EET with 0.05 percent N-lauroylsarcosine [Sigma, St. Louis] and 200 µg of lysozyme per milliliter [Boehringer Mannheim, Indianapolis]) for four hours at 30°C, followed by proteinase K (EET with 1.0 percent sodium dodecyl sulfate and 1 mg of proteinase K per milliliter; Boehringer Mannheim) for at least six hours at 50°C and were washed four times with TE buffer (10 mM Tris, pH 8.0, and 1 mM EDTA) for 30 minutes. All the washing steps were conducted with rotation at 170 cycles per minute by using a 1.0-in.-diameter orbital stroke. Inserts were cut into lengthwise slices and restricted with Xba1 endonuclease in 100 µl of buffer H (Boehringer Mannheim; 30 units) for at least four hours at 37°C. Restricted slices were washed in 0.5x TBE (which consists of 5.4 g of Tris base per liter, 2.75 g of boric acid per liter, and 0.645 g of EDTA per liter) and loaded directly onto a 1 percent agarose gel (FastLane, FMC Bioproducts).
PFGE Conditions
DNA fragments were separated with a CHEF Mapper (Bio-Rad Laboratories, Hercules, Calif.). Lambda concatemers and Saccharomyces cerevisiae DNA (Boehringer Mannheim) were used as standards. Gels were run with the use of 0.5x TBE buffer at 14°C, a linear ramp of 12.56 to 40.09 seconds over a period of 24 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) and photographed with ultraviolet illumination from a fixed camera position.
Gel Analysis and Interpretation
Restriction-fragment patterns were initially matched by visual comparison and measurement of physical band migration relative to lambda and yeast standards. Gel photographs were subsequently digitized with a Photodoc 1000 System (Bio-Rad), and PFGE profiles were compared by using the Jaccard coefficient and Molecular Analyst Plus software (Bio-Rad). DNA patterns were screened with a 3 percent molecular weight–matching tolerance, and matches were confirmed visually. PFGE strains were defined as having exact matches of all bands in the 147-to-700-kb range. A single isolate representing each PFGE pattern was retested to verify its uniqueness and to establish run-to-run reproducibility. On each gel, an E. coli O157:H7 reference strain was used in three lanes as a process control (Figure 1).
|
Three hundred forty-four cases of E. coli O157:H7 infection were reported among Minnesota residents during 1994 (147 cases) and 1995 (197). The annual rate of reported infections among Minnesota residents over the two-year period was 3.8 per 100,000 person-years (Table 1). The age-specific rate of infection for children under the age of five years was 9.3 per 100,000 person-years.
|
|
|
|
|
Of the 10 outbreaks and 35 case clusters, 9 involved five or more isolates of an identical subtype pattern; a common source was identified in 6 (67 percent) of those. Conversely, of the 36 clusters involving two to four isolates with a single PFGE pattern, a common source was identified in only 4 (11 percent).
A high level of homogeneity was noted among isolates obtained from epidemiologically linked cases. Of the 56 cases identified as part of the 10 outbreaks recognized in 1994 and 1995, only 2 cases in 2 different outbreaks involved a PFGE pattern different from the recognized outbreak strain. Although these isolates differed from the outbreak strain by one band, they were closely related to it when evaluated according to the interpretive criteria proposed by Tenover et al.10
Twenty-five cases of the hemolytic–uremic syndrome were identified during the two-year period — 9 in 1994 and 16 in 1995. Fourteen patients (56 percent) had E. coli O157:H7 isolated from stool specimens. Eleven different E. coli O157:H7 patterns were recognized among these patients. No enteric pathogen was isolated from the other 11 patients. Twenty-two patients with the hemolytic–uremic syndrome (88 percent) were under 10 years of age, and 76 percent (19) were male. The annual age-specific rate of the hemolytic–uremic syndrome in children less than 10 years of age with documented E. coli O157:H7 infection was 1.2 per 100,000 person-years. Fifteen cases of the syndrome (60 percent) occurred in July, August, and September, reflecting the same seasonal trend as was documented for E. coli O157:H7 infections overall. The ratio of E. coli O157:H7 infections to cases of the hemolytic–uremic syndrome among children less than 10 years of age did not vary significantly according to the geographic area within the state. In the Minneapolis–St. Paul metropolitan area, there were 4.3 E. coli O157:H7 infections reported for each case of the hemolytic–uremic syndrome. Outside the metropolitan area, 5.2 cases of E. coli O157:H7 infection were reported for each case of the syndrome.
Discussion
We report here the results of molecular subtyping of E. coli O157:H7 isolates and the usefulness of this method in public health practice. Our major findings include the following. First, there was a marked heterogeneity of isolates, with 143 distinct PFGE patterns identified during the two-year study period. Second, subtyping provided a tool for determining whether or not an outbreak or multiple outbreaks were present when the number of cases increased. For most of the sudden increases in numbers of reported cases, there was no evidence of an outbreak when isolates were subtyped. Without the benefit of subtyping, investigations of similar increases in other areas of the country have led to unnecessary use of limited public health resources.8,11 In two instances in Minnesota, what originally appeared to be a single outbreak was subsequently recognized through subtyping to be multiple outbreaks occurring within a defined period. Third, subtype-specific surveillance allowed us to identify outbreaks that would not have been detected by traditional surveillance methods. Almost half the outbreaks documented (4 of 10) were identified by subtype-specific surveillance alone.
The heterogeneity of the E. coli O157:H7 patterns identified in our study has implications for efforts to prevent and control infections. Since most case clusters were small and many cases were sporadic, primary-prevention strategies must focus on reducing contamination of raw foods and changing behavior, particularly behavior related to food preparation and general hygiene.1,12,13
During the two-year period, over half the patients identified harbored E. coli isolates that had unique PFGE patterns. However, 30 percent of the patients had an E. coli isolate with a PFGE pattern that was isolated from at least one other person during the same two-week period but for which no epidemiologic association could be identified. Because cases involving E. coli isolates with similar PFGE patterns often clustered temporally, the data suggest that different E. coli strains may contaminate a common product and subsequently appear in the community where the product is distributed. Such outbreaks go unrecognized because relatively few cases are detected. Even if molecular subtyping is used on a statewide basis and multiple isolates with common PFGE patterns are identified, it may be difficult to determine the source if the numbers of cases are low, unless amplification occurs through other community exposures, such as the contamination of swimming beaches or transmission within child-care settings.14,15,16,17 For example, when a cluster of five or more cases of a single PFGE pattern was recognized, we found a common source 67 percent of the time. In contrast, when only two to four cases with the same pattern were identified, a common source was found only 11 percent of the time. These results suggest that the identification of outbreaks is unlikely when commercial products are distributed over a multistate area and few cases are identified in each state. Active subtype surveillance on a regional basis may be necessary to increase the likelihood of detecting such outbreaks, identifying the source, and taking appropriate public health action.
The large number of PFGE patterns found in this study can be attributed in part to the stringency of the definition of a strain. Strain relatedness, determined according to the interpretive criteria proposed by Tenover et al.,10 was not used to differentiate among PFGE patterns in this population-based study. As Tenover et al. state, their criteria for determining strain relatedness were designed for the evaluation of outbreaks of limited size and duration. Conversely, we believe that the interpretive criteria for strain matching of isolates obtained through population-based surveillance should be more restrictive to provide the specificity needed to relate apparently sporadic infections over a large geographic area. This approach is critical to the detection of outbreaks associated with widely distributed food products. Furthermore, our data indicate that PFGE patterns tend to be stable in outbreaks, although single-band differences were found in two instances among isolates from epidemiologically associated cases.
Using PFGE subtyping (or other subtyping methods as they become available) as a routine surveillance tool requires, first, a system for submitting and receiving isolates. In Minnesota, clinical laboratories are required to notify the Minnesota Department of Health of any pathogenic E. coli isolated within 24 hours of the completion of microbiologic cultures and to submit isolates to the state public health laboratory. Second, laboratory staff must be available at a central location to characterize isolates rapidly as they are submitted. Generally, results were available within five days when there was one full-time PFGE technician. It is also essential to have the epidemiologic capacity to investigate cases rapidly when identical PFGE patterns are identified. Although this system may seem expensive, much of the cost can be offset by not initiating epidemiologic investigations when the number of cases increases unless subtyping data support the possibility of a common source. Increasing the specificity of outbreak case definitions by the use of molecular subtyping can conserve resources by decreasing the amount of time spent on unnecessary and labor-intensive case–control studies.
Molecular subtyping of E. coli O157:H7 has revolutionized population-based surveillance for this organism in Minnesota. We now routinely subtype all E. coli O157:H7 isolates and consider this technique to be an integral part of disease prevention and control in our state. We are currently exploring similar subtyping approaches for other enteric and invasive bacterial pathogens, and we believe that such techniques will substantially improve other public health surveillance efforts, particularly at the state and regional levels.
Supported in part by a cooperative agreement (U50/CCU511190) with the CDC.
Source Information
From the Acute Disease Epidemiology Section (J.B.B., C.W.H., K.L.M., M.T.O.) and the Division of Public Health Laboratories (J.M.B., D.J.B.), Minnesota Department of Health, Minneapolis.
Address reprint requests to Dr. Bender at the Acute Disease Epidemiology Section, Minnesota Department of Health, 717 Delaware St. SE, Minneapolis, MN 55440-9441.
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
