Routine stool cultures do not identify E. coli O157:H7. Unlike 80% of E. coli serotypes, however, E. coli O157:H7 does not rapidly ferment D-sorbitol and therefore appears colorless on sorbitol-MacConkey agar culture plates read at 24 hours [15, 16]. These sorbitol-negative colonies can then be screened for agglutination in O157 antiserum.

Relatively little information is available about the frequency of isolation of E. coli O157:H7 from ill persons in the United States; most U.S. laboratories do not routinely culture for this organism [17]. In single-center studies in the United States in which all stool specimens were cultured for this organism, isolation rates ranged from 0.08% to 0.5% [18-20]. Recent information suggests that E. coli O157:H7 has been isolated from patients in most states [21], but the frequency of this isolation compared with that of other enteric pathogens in different geographic areas during similar time periods has not been described.

The reported clinical signs and symptoms of E. coli O157:H7 infection include bloody or nonbloody diarrhea, abdominal cramps, and lack of reported fever [22, 23]. However, this information was derived from relatively few persons in outbreak settings, and few studies have examined the clinical presentation of illness due to E. coli O157:H7 infection compared with the clinical presentation of illness due to other bacterial enteric pathogens.

We therefore sought to determine 1) the frequency of isolation of E. coli O157:H7 and 2) the clinical and epidemiologic features of infections with E. coli O157:H7 compared with those of Campylobacter, Salmonella, and Shigella species at 10 hospitals located throughout the United States. Using standard microbiological methods, we assessed the ways in which time of year, geographic location, and the demographic and clinical features of patients affected the likelihood of isolation of these enteric pathogens.

Methods

Top Methods Results Discussion Author & Article Info References

Study Sample

Our study was announced and participation was requested in a newsletter that was sent to hospitals in the National Nosocomial Infections Surveillance system [24]. Five hospitals in this system and five other hospitals were chosen on the basis of geographic location, willingness to participate, receipt of specimens in a primary care setting, and the expectation that an adequate number of outpatient stool cultures would be done each year. All four census divisions of the United States were represented. Nine hospitals served general patient populations that included all age groups, and one served a primarily pediatric population. All served both inpatients and outpatients. The average annual number of stool specimens screened by each hospital ranged from 400 to 4000 (median, 1300). Four of the hospitals were university hospitals, and six were community hospitals.

At each hospital, all of the specimens studied were fecal samples from inpatients and outpatients of all ages that were submitted to the clinical microbiology laboratory for routine pathogen identification. The study was conducted from October 1990 through October 1992.

Collection and Handling of Specimens

All sites agreed to use the following methods for the collection and handling of specimens. Swabs were transported in Cary-Blair transport medium, Amies transport medium, or Stuart transport medium and were streaked immediately onto plating media or were kept at 4 °C for no more than 24 hours. If whole stool specimens were not examined within 1 hour of receipt by the laboratory, a swab of the stool was placed in transport medium, refrigerated, and examined within 24 hours. Specimens were visually inspected for gross blood, and the presence of occult blood was determined by using the hemoccult test. The presence of fecal leukocytes was determined by placing a bit of stool in a drop of methylene blue on a slide or by doing a Gram stain and examining the specimen using the high-power microscope objective. Specimens were graded as having 0, 1 to 4, 5 to 9, or 10 or more leukocytes per high-power field. Standard methods were used to isolate and identify Campylobacter, Salmonella, and Shigella species. Other assays, such as those for Clostridium difficile or rotavirus, were not part of the protocol and were done according to the routines of the individual site laboratories and the physicians ordering the tests.

Isolation of Escherichia coli O157:H7

Before the study began, each laboratory received control strains of E. coli O157:H7 and instructions about the isolation and identification of this organism. To identify E. coli O157:H7, fecal specimens were plated onto sorbitol-MacConkey agar and the plates were incubated at 37 °C for 24 hours. Three sorbitol-negative colonies were tested for agglutination with O157 latex reagents (Pro-Lab, Inc., Round Rock, Texas). The O157-positive colonies were sent to the Centers for Disease Control and Prevention for biochemical identification and serotyping [25]. Isolates confirmed as E. coli O157:H7 or O157:NM (nonmotile) were tested for production of Shiga toxin 1 and 2 (formerly called Shiga-like toxins I and II [26]) and for the presence of Shiga toxin genes by hybridization with oligonucleotide probes [27]. Isolates were tested by using the disk diffusion technique [28] for susceptibility to a standard panel of antimicrobial agents [28].

Data Collection

For each fecal specimen received, data were entered on a standard line list; only the first specimen from each patient was included. The information collected for each specimen included date obtained, date plated, source (whole stool or swab), presence of visible or occult blood, presence and quantity of fecal leukocytes, and presence of pathogens.

After permission was obtained from the relevant health care provider, a clinical data form was completed through retrospective chart review for all patients from whom Campylobacter, Salmonella, or Shigella species or E. coli O157:H7 were isolated and from every 25th patient from whom no pathogen was isolated. Information obtained included age, sex, date of the onset of illness, inpatient or outpatient status, symptoms (including presence and date of onset of diarrhea, bloody stools, abdominal pain, vomiting, and fever in the previous 2 weeks), abdominal tenderness, largest number of bowel movements in a 24-hour period, maximum body temperature on the day of culture (as measured by a health practitioner), peripheral blood leukocyte count, and whether the patient was admitted to the hospital. If a patient had a body temperature of at least 37.8 °C, he or she was considered to have fever.

Data Analysis

Salmonella, Shigella, and Campylobacter species and E. coli O157:H7 were considered to be major bacterial enteric pathogens. Isolates that were identified as O157:H7 or O157:NM and that produced Shiga toxin were considered to be strains of E. coli O157:H7. An isolation proportion for a pathogen was defined as the proportion of all fecal specimens that yielded that pathogen. To estimate the age-specific isolation proportion of pathogens from stool specimens, we divided the number of persons in each age group for whom a specific pathogen was isolated by the sum of all persons (both culture-positive and culture-negative) in that age group. We estimated the total number of culture-negative persons in each age group by extrapolating the age group distribution frequency from the sample of culture-negative persons for whom age was known to all culture-negative persons. For the analysis of clinical features associated with infection, patients whose stool cultures yielded more than one bacterial pathogen were excluded.

Differences in proportions were analyzed using a chi-square test or the Fisher exact test. For normally distributed data, differences in means were compared using the Student t-test; for nonparametric data, differences in medians were compared using the Wilcoxon two-sample test. Logistic regression analysis was done using generalized estimating equations to assess factors independently associated with E. coli O157:H7 infection while controlling for study site. For all statistical tests, a two-tailed P value less than 0.05 was considered significant.

Results

Top Methods Results Discussion Author & Article Info References

Isolation of Pathogens

During the study period, fecal specimens from 30 463 persons were examined. A source was specified for 29 355 of these specimens; 63% were from whole stools and 37% were from swabs. Overall, 1708 of the specimens (5.6%) yielded at least one of the four major bacterial enteric pathogens; for 27 902 specimens (91.6%), no pathogen was isolated. Eleven patients had dual infections: Six had Shigella species and Campylobacter species infections, 3 had Salmonella species and Campylobacter species infections, and 2 had Shigella species and Salmonella species infections.

The highest isolation proportions from fecal specimens for E. coli O157:H7 were seen in hospitals in Maine and Wisconsin; the lowest proportion was seen in Virginia (Table 1). Of the four bacterial pathogens, E. coli O157:H7 was the second most frequently isolated in Maine, the third most frequently isolated (ahead of Shigella species) in Washington and Wisconsin, and the third most frequently isolated (tied with Shigella species) in Michigan. In the hospitals in northern states (Maine, Michigan, New York, Utah, Washington, and Wisconsin), the isolation proportion for E. coli O157:H7 was 0.57%; in the hospitals in southern states, it was 0.12% (P < 0.001). Isolation proportions for E. coli O157:H7 did not differ significantly between hospitals in the eastern and western United States (0.36% compared with 0.42%).

Escherichia coli O157:H7 was isolated most frequently during the summer months. Of 116 specimens that yielded E. coli O157:H7 and for which the date of collection was known, 74 (63.8%) were collected from June through September. The isolation of Campylobacter species and, to a lesser extent, Salmonella species also peaked during the summer months. Isolation proportions for Shigella species were higher between October and December.

The presence of any fecal leukocytes or the presence of at least 10 fecal leukocytes per high-power field was seen significantly more often in specimens yielding E. coli O157:H7 than in specimens yielding the other pathogens (P < 0.001) (Table 2). In a stratified analysis that accounted for the presence of visible blood, specimens yielding E. coli O157:H7 were still significantly more likely than specimens yielding other bacterial enteric pathogens to have fecal leukocytes (P = 0.001).

Information on the presence of visible blood was available for 71.6% of specimens. Of the 658 specimens (3.0%) that were visibly bloody, 132 (20.1%) yielded one of the four major bacterial pathogens. In addition, 24 of the 658 specimens (3.7%) were positive for C. difficile toxin and 5 (0.8%) were positive for other organisms (3 for rotavirus, 1 for Yersinia species, and 1 for Aeromonas species). Isolation of E. coli O157:H7 was strongly associated with the presence of visible blood in the stool. Of 658 stool specimens with visible blood, 51 (7.75%) yielded E. coli O157:H7; of the 21 164 specimens without visible blood, 30 (0.14%) yielded E. coli O157:H7 (odds ratio [OR], 59.2 [95% CI, 36.6 to 96.0]).

Overall, 63.0% of the 81 E. coli O157:H7 isolates for which information was available came from visibly bloody specimens; corresponding values for the other pathogens were 14.7% of 170 for Shigella species, 7.8% of 472 for Campylobacter species, and 4.8% of 400 for Salmonella species. The pathogen most commonly isolated from visibly bloody stool specimens was E. coli O157:H7: It accounted for 39% (median, 20%; range, 0% to 100% across sites) of specimens that yielded a major bacterial enteric pathogen, and it was the pathogen first or second most commonly isolated from visibly bloody stool specimens at 6 of the 10 study sites. Among the hospitals in southern states, E. coli O157:H7 was the pathogen third or fourth most commonly isolated from visibly bloody stool specimens; it was found in a median of 10% (range, 0% to 14%) of specimens that yielded a major bacterial enteric pathogen. Of the 64 E. coli O157:H7 specimens that underwent hemoccult testing, 82.8% had positive results. Among the specimens that had other bacterial enteric pathogens, the proportion that were positive on hemoccult testing was 59.1% of 105 for Shigella species, 52.0% of 348 for Campylobacter species, and 43.4% of 198 for Salmonella species.

The largest number of E. coli O157:H7 isolates was obtained from children 1 to 4 years of age (18 specimens) and adults 60 to 69 years of age (17 specimens). Age-specific isolation proportions from fecal specimens were highest among patients 5 to 9 years of age (9 specimens; 0.90%) and 50 to 59 years of age (13 specimens; 0.89%); the lowest age-specific isolation proportion was among infants (2 specimens; 0.06%). Compared with the other major bacterial enteric pathogens, E. coli O157:H7 had the lowest age-specific isolation proportion among persons younger than 50 years of age, both overall and within each age group. Among persons older than 50 years of age, the isolation proportion was higher for E. coli O157:H7 than for Shigella species (0.36% compared with 0.24%). The isolation proportion for E. coli O157:H7 was not significantly different for male and female patients.

All 118 E. coli O157:H7 isolates produced at least one Shiga toxin: One hundred two (86.4%) produced Shiga toxin 1 and Shiga toxin 2, 14 (11.9%) produced Shiga toxin 2 alone, and 2 (1.7%) produced Shiga toxin 1 alone. No geographic region was associated with a particular toxin type. Six E. coli O157 isolates were NM; of these, 4 were identified in Wisconsin, 1 was identified in New York, and 1 was identified in Washington.

Ten (8.5%) of 118 E. coli O157:H7 isolates were resistant to at least one antimicrobial agent; no geographical clustering was seen. Of these 10 isolates, 3 were resistant to one antimicrobial agent (1 to ampicillin, 1 to tetracycline, and 1 to streptomycin). Of the 7 isolates that showed resistance to more than one antimicrobial agent, 3 were resistant to tetracycline, sulfisoxazole, and streptomycin; 1 was resistant to sulfisoxazole and chloramphenicol and had intermediate resistance to streptomycin; 1 was resistant to sulfisoxazole, streptomycin, and ampicillin; 1 was resistant to tetracycline, sulfisoxazole, streptomycin, and gentamicin; and 1 was resistant to sulfisoxazole, streptomycin, ampicillin, and trimethoprim-sulfamethoxazole.

Clinical Features of Infected Patients

Diarrhea was a reported symptom in 96% of culture-positive patients and 83% of culture-negative patients. The prevalence of clinical signs and symptoms among patients with one of the four major bacterial enteric pathogens is shown in Table 2. In univariate analysis, patients with E. coli O157:H7 infection were significantly more likely than patients infected by another bacterial enteric pathogen to report a history of bloody diarrhea (OR, 16.6 [CI, 8.0 to 35.6]) or abdominal cramps (OR, 3.0 [CI, 1.4 to 6.4]); fever was reported significantly less often (OR, 0.26 [CI, 0.17 to 0.42]). Physical and laboratory findings that were significantly more frequent in patients with E. coli O157:H7 infection than in patients infected with the other bacterial enteric pathogens included abdominal tenderness (OR, 4.2 [CI, 2.6 to 7.0]), visible blood in the stool specimens (OR, 20.4 [CI, 11.9 to 34.9]), hemoccult-positive stool specimens (OR, 4.7 [CI, 2.3 to 9.7]), the presence of any fecal leukocytes (OR, 4.0 [CI, 2.4 to 6.6]), and the presence of at least 10 fecal leukocytes per high-power field (OR, 2.1 [CI, 1.2 to 3.6]).

In a logistic regression model, clinical signs or symptoms independently associated with E. coli O157:H7 infection compared with Campylobacter, Salmonella, or Shigella species infection included reported bloody diarrhea (OR, 18.6 [CI, 7.4 to 48.6]), visibly bloody stool specimens (OR, 8.1 [CI, 3.6 to 18.3]), no reported fever (OR, 8.3 [CI, 1.6 to 50.0]), a peripheral leukocyte count greater than 10 x 10⁹/L (OR, 4.0 [CI, 1.7 to 9.5]), and abdominal tenderness (OR, 2.9 [CI, 1.2 to 7.2]). At least three of these features were present in 65.4% of patients with E. coli O157:H7 infection; they were present in 18.5% of those who had infection with Campylobacter, Salmonella, or Shigella species (P < 0.001).

In a logistic regression model that assessed clinical and laboratory findings among patients whose stool specimens yielded a major bacterial pathogen and were not visibly bloody, only a history of bloody diarrhea (OR, 12.5 [CI, 7.2 to 21.7]) or abdominal tenderness on physical examination (OR, 1.8 [CI, 1.1 to 3.0]) was associated with E. coli O157:H7 infection.

Table 3 shows the likelihood of the presence of any particular clinical feature or laboratory finding among the patients with one of the four bacterial enteric pathogens compared with the randomly selected culture-negative comparison group. In univariate analysis, patients infected with any of the four major bacterial pathogens had all clinical features (except for elevated peripheral leukocyte count and objective fever) significantly more frequently than did patients in the culture-negative comparison group. Patients with E. coli O157:H7 infection were significantly less likely to have objective fever than did patients in the culture-negative control group, and they were more likely to have an elevated peripheral leukocyte count.

Screening Efficiency

To help clinicians and laboratory personnel use microbiology resources efficiently, we examined how well various clinical symptoms or signs served as criteria to indicate when a stool specimen should be cultured for E. coli O157:H7. Efficiency of the criteria was measured as the proportion of all E. coli O157:H7 infections that were detected divided by the proportion of all stool specimens that would have been cultured using these selected criteria (Table 4). Among single criteria, visible blood in the stool specimens was by far the most efficient: When this criterion alone was used, only 3% of stool specimens would need to be cultured to detect 63% of all E. coli O157:H7 infections. However, use of this criterion alone would not detect one third of infections. If either the visible blood in the stool specimen seen in the laboratory or a history of bloody diarrhea obtained by the clinician was used as a criterion, more than 90% of E. coli O157:H7 infections would be detected by screening only 22% of all stool specimens. Other criteria in combination with the presence of visible blood were less efficient.

Discussion

Top Methods Results Discussion Author & Article Info References

The results of our 10-hospital survey underscore the importance of E. coli O157:H7 as a cause of bacterial diarrhea in the United States. The overall isolation proportion of 0.39% was more than one third that of Shigella species; in some geographic areas and age groups, E. coli O157:H7 surpassed some of the other routinely sought bacterial enteric pathogens in frequency of isolation. The E. coli O157:H7 isolation proportion in our study was similar to that seen in smaller, single-center studies in Seattle (0.4%) [18] and Rochester, Minnesota (0.5%) [19], and was higher than that seen in Chicago (0.08%) [20].

Although our study was larger and geographically more diverse than other studies that have examined the incidence of E. coli O157:H7 in the United States, our methods had certain limitations. The hospitals selected for the survey represented all of the geographic areas of the United States but were not chosen at random; thus, the populations served by these institutions may not have been representative of the general population of the United States. Because this was a hospital-based study rather than a population-based study, the age distribution and general health of the study participants were not representative of the general population. Stool cultures were obtained through decisions by individual physicians rather than for all patients whose symptoms met a uniform definition of diarrhea; patients who had cultures were likely to be more ill than all patients with diarrhea seeking medical care; and specimen characteristics were representative of patients for whom a decision to obtain a stool culture had already been made. Therefore, extrapolations from our results to all patients with diarrhea who seek medical care must be made cautiously. Finally, the proportion of cultures with certain characteristics that yielded particular pathogens represent isolation proportions from fecal specimens, not population-based incidence rates. Isolation proportions may depend on many different factors, including physician practices with regard to stool culturing and the prevalence of various pathogens among different age groups. Thus, they may not be directly comparable to population-based rates or risks. Nonetheless, the frequency of isolation for E. coli O157:H7 in our study was similar to that in previous studies, as were the isolation proportions from fecal specimens for other enteric pathogens [29].

Although E. coli O157:H7 was found in all geographic areas, isolation proportions varied from 0.04% to 2.13% among study sites, and the organism was among the three most frequently isolated bacterial enteric pathogens at 4 of the 10 sites. Escherichia coli O157:H7 was isolated significantly more frequently at northern than at southern sites. This finding is consistent with stool survey studies done in Canada, which have documented even higher E. coli O157:H7 isolation proportions than have been found in the United States [30, 31]. In the United States, both sporadic cases and outbreaks of E. coli O157:H7 are reported more frequently in northern than in southern states [32]. The reasons for this variation are unknown, but geographic differences in animal carriage, meat processing methods, or consumption of or cooking methods for ground beef are possibilities [33, 34]. The practices of physicians with regard to stool culturing may also vary by region.

Most E. coli O157:H7 organisms were isolated during the warmer months of the year; almost two thirds were isolated from June through September. A similar seasonal variation has been seen in other studies [18, 35]. The reasons for this variation are unknown; however, a study in Washington state [36] found that in cattle, fecal shedding of E. coli O157:H7 increased during the summer months. Seasonal changes in such factors as handling, cooking, or patterns of consumption of ground beef may contribute to these differences.

In our study, the isolation proportion of E. coli O157:H7 from fecal specimens in adults 50 to 59 years of age was the second highest among any age group, and E. coli O157:H7 was isolated more frequently than Shigella species among adults 50 years of age or older. Previous population-based studies [18, 31, 35] have documented relatively lower rates among adults than among children, and data from outbreaks in day care centers have shown that the youngest children have the highest risk for infection [13, 14]. Age-specific isolation proportions from fecal specimens are not equivalent to population-based incidence rates; however, our data suggest that many E. coli O157:H7 infections are seen in adults 50 years of age or older. This suggests that screening for E. coli O157:H7 only in specimens from children, a strategy used by some clinical laboratories, may miss many infections [17].

Because E. coli O157:H7 was the bacterial pathogen most commonly isolated from visibly bloody stool specimens, clinical laboratories should, at a minimum, culture all visibly bloody stool specimens for this organism. However, a patient history of bloody diarrhea was a more sensitive indicator of E. coli O157:H7 infection than was the presence of visibly bloody stool specimens. Although more than 90% of our patients with E. coli O157:H7 infection reported a history of bloody diarrhea, only 63% of those for whom information was available had a specimen noted as visibly bloody in the laboratory. In addition, information on visible blood was not available for two thirds of specimens taken from swabs. Because person-to-person transmission of E. coli O157:H7 is not uncommon and because infection with this organism may result in severe disease, we recommend that stool specimens from all patients with a history of acute bloody diarrhea be cultured for E. coli O157:H7.

Others [37] have recommended that all stool specimens submitted for examination for bacterial enteric pathogens be cultured for E. coli O157:H7. This approach is reasonable because, in some age groups or at some sites, we found E. coli O157:H7 to be more common than Shigella species. Overall, E. coli O157:H7 was isolated one third as often as Shigella species, a routinely sought bacterial enteric pathogen.

To save money, some clinical laboratories choose not to screen all submitted stool specimens but rather use various criteria to identify specimens that should be cultured for E. coli O157:H7 [17]. Using clinical information in combination with specimen characteristics available to the microbiologist can improve sensitivity while providing reasonable efficiency. In our study, supplementing the presence of visible blood in the specimen with a patient history of bloody diarrhea increased the sensitivity of detection over that achieved by the presence of visible blood alone and was the most efficient of those approaches that detect more than 90% of E. coli O157:H7 infections. These proportions and ratios should be viewed with caution because our study was not designed to directly address screening efficiency, and decisions to do stool cultures were not made prospectively on the basis of patient symptoms or specimen characteristics. Nonetheless, our data suggest that providing clinical information to laboratories, perhaps as a specific notation or "check box" on the laboratory request, can improve laboratory efficiency and focus diagnostic efforts. As laboratories develop "rejection criteria" for clinical specimens, the exchange of information between clinicians and laboratory personnel will be increasingly important [38].

More patients with E. coli O157:H7 had either any fecal leukocytes or at least 10 fecal leukocytes per high-power field than did patients with the other bacterial pathogens. In contrast, reports from smaller studies [22] have suggested that fecal leukocytes may be infrequent or scanty in patients with this infection. Our results suggest that fecal leukocytes in a stool specimen should not dissuade the clinician from considering the diagnosis of E. coli O157:H7 infection; in fact, this finding may increase the likelihood of such infection.

The proportion of patients with E. coli O157:H7 infection who have abdominal tenderness has not been frequently reported. In our study, most patients with this infection had abdominal tenderness, and the symptom occurred more often in patients with E. coli O157:H7 infection than in persons with other bacterial enteric infections. Other investigators [10, 39, 40] have documented evidence of right-sided colonic inflammation by barium enema or colonoscopy in patients with this infection. Abdominal tenderness associated with E. coli O157:H7 infection may contribute to misdiagnoses, such as appendicitis or intussusception, and may result in unnecessary surgical procedures [2].

The cause of diarrhea among the many patients in our study who had negative stool cultures remains an intriguing unknown. Some of these patients may have had a bacterial infection and received antimicrobial therapy before cultures were obtained. Because information on the previous use of antimicrobial agents was not available, we could not assess this possibility. The length of time between onset of symptoms and culture may also affect the likelihood of a positive stool culture. In our study, the median duration of illness before culture was significantly longer in the culture-negative group than in those from whom a bacterial pathogen was isolated (4 compared with 3 days; P < 0.001). Some patients with negative cultures may have had chronic diarrhea as a result of inflammatory bowel disease or other conditions. It is also likely that some of the culture-negative patients were infected with as-yet unrecognized pathogens or pathogens that are not routinely sought by typical clinical laboratories, such as enterotoxigenic E. coli [41]. In this regard, a particularly interesting subset is the 75% of patients with visibly bloody stool specimens for whom no pathogen was identified. An outbreak of bloody diarrhea due to a non-O157 Shiga toxin-producing E. coli serotype, O104:H21, was recently recognized in Montana [42], and Shiga toxin-producing E. coli O111:NM was linked to a large outbreak of bloody diarrhea and the hemolytic uremic syndrome in Australia [43]. Some cases of culture-negative bloody diarrhea may be due to these and other non-O157 Shiga toxin-producing E. coli; these organisms can be detected by examining stool specimens for the presence of Shiga toxin or Shiga toxin-producing organisms using enzyme-linked immunoassay, polymerase chain reaction, or genetic probes. Clusters of persons with bloody diarrhea or the hemolytic uremic syndrome whose stool cultures do not yield E. coli O157 or other pathogens should prompt providers to request state health department laboratories to examine specimens for other Shiga toxin-producing E. coli.

Appendix

The following are the Escherichia coli O157:H7 Study Group investigators: Yoginder Chitkara, MD, and Kathy McCasland (St. Mary's Hospital, Tucson, Arizona); Stuart Cohen, MD, and Elizabeth A. Beckley (University of California Davis Medical Center, Sacramento, California); Stephen G. Jenkins, PhD, and Joan Abid (Baptist Medical Center, Jacksonville, Florida); Sara Whelan, MS, and Margaret Ostrander (Mercy Hospital, Portland, Maine); Shirley L. Jackson (Port Huron Hospital, Port Huron, Michigan); Marjorie Berman, MS (North Shore University Hospital, Manhasset, New York); Andrew T. Pavia, MD, and Judy Daly, PhD (University of Utah, Salt Lake City, Utah); Richard Duma, MD, and Harry Dalton, PhD (Medical College of Virginia, Richmond, Virginia); Donald E. Anderson, PhD, and Patricia Bauer (Sacred Heart Medical Center, Spokane, Washington); and Ellen Smith, RN, and Richard Dern, MS (St. Mary's Hospital Medical Center, Madison, Wisconsin).

From the Centers for Disease Control and Prevention, Atlanta, Georgia.

Ms. Hutwagner: Biostatistics and Information Management Branch, National Center for Infectious Diseases, Centers for Disease Control and Prevention, Atlanta, GA 30333.