Abstract
This study describes the performance of two rapid enzyme immunoassays, Premier E. coli O157 and Premier EHEC (Meridian Diagnostics Inc., Cincinnati, Ohio) for the detection in stools of Escherichia coli O157 and verotoxins (Shiga-like toxins), respectively. Both tests were performed on stools from 876 children presenting to eight emergency departments with diarrhea. Standard culture, including E. coli O157:H7 isolation, was performed, and paired sera were taken for anti-O157-lipopolysaccharide antibody determination. Stools from patients enrolled in the study, and those yielding discordant results, were sent to a reference laboratory for repeat testing and further investigation, including cytotoxicity and non-O157 verotoxin-producing E. coli culture. Results were classified as field results (obtained in the eight site laboratories) and resolved results (obtained after repeat testing in the central laboratory). The “gold standard” for sensitivity of both tests and for specificity of Premier E. coli O157 was isolation of E. coli O157:H7 or a fourfold anti-O157 antibody rise. Specimens positive by the Premier EHEC test and negative for E. coli O157 culture were examined for non-O157 verotoxin-producing E. coli. The field sensitivity of Premier E. coli O157 was 86%, that of Premier EHEC was 89%, and the specificity of Premier E. coli O157 was 98%. Ten of 13 discordant Premier E. coli O157 results were reassigned as true results after repeat testing. Ten non-O157 verotoxin-producing E. coli isolates were recovered from Premier EHEC-positive, E. coli O157 culture-negative stools. Only one specimen gave an unequivocally false-positive Premier EHEC result. Both tests are highly sensitive and are specific if correctly performed. The Premier EHEC test will be particularly valuable as a practical routine test for the detection of non-O157 verotoxin-producing E. coli.
Approximately 8 to 10% of children with diarrhea caused by Escherichia coli strains which are capable of producing verotoxins (VTs) (Shiga-like toxins) develop hemolytic-uremic syndrome (HUS) 5 to 10 days after the onset of symptoms (4). Antigenically distinct verotoxins (VT1 and VT2) can cause endothelial injury in vitro. Because endothelial injury is thought to precipitate the microangiopathy that results in HUS, it has been hypothesized that HUS results from endothelial injury caused by systemic spread of these toxins (6, 8–10). No intervention has proven effective in preventing HUS. The efficacy of an oral synthetic toxin receptor analog (SYNSORB Pk) in preventing systemic spread of VT, and subsequent development of HUS, is currently being evaluated in a phase 3 randomized controlled trial. In an earlier phase 2 trial of the same agent, patients were recruited by clinical criteria in order to begin treatment as early as possible in the disease process. Once enrolled, many had to be excluded from the analysis of the efficacy of the intervention because laboratory tests did not confirm the presence of verotoxin-producing E. coli (VTEC). To improve the efficiency of enrolling children with true VTEC infection in the phase 3 trial, two enzyme immunoassays (EIAs) were performed on all potentially eligible subjects with diarrhea. The first test (Premier E. coli O157) detects the presence of O157 antigen in stool, and the second (Premier EHEC) detects the presence of VT1 and VT2. One difficulty with the clinical evaluation of the Premier EHEC test is that the logical “gold standard” would be the stool cytotoxicity test, but this test is not widely available, and specimen transport under appropriate conditions to a laboratory that can perform it may be difficult to organize, so that the sensitivity of the enterohemorrhagic E. coli (EHEC) test has to be evaluated by using E. coli O157:H7 isolation as a surrogate gold standard. This paper reports our experience with these tests and partially addresses this problem.
MATERIALS AND METHODS
Design of study.
The Premier E. coli O157 and the Premier EHEC tests (Meridian Diagnostics Inc., Cincinnati, Ohio) were performed on stool specimens obtained from 876 children between the ages of 1 month and 8 years presenting with acute diarrhea to emergency departments in eight centers (sites) in Canada between 1 June 1996 and 31 October 1996. Stools were also cultured for E. coli O157:H7 and for other bacterial stool pathogens, each center using its own routine procedures. All sites used sorbitol-MacConkey agar for the isolation of E. coli O157:H7. These tests were performed as part of the screening procedure for selection of patients for enrollment in a phase 3 therapeutic trial of SYNSORB Pk for the prevention of HUS following VTEC infection. One center (Hôpital Sainte-Justine, Montréal, Canada) performed the two EIAs on all stools submitted to the laboratory for the study period, regardless of age or eligibility for the therapeutic trial. One of the eight sites performed only a direct Premier EHEC without the incubation step (see “EIAs”). The results from this site were omitted from the Premier EHEC calculation of sensitivity.
Paired sera were taken on days 1 and 8 from children enrolled in the clinical trial for measurement of E. coli O157 lipopolysaccharide antibody response.
Stool specimens from patients enrolled in the study and those giving discordant results were transported, frozen at −56°C, to the reference laboratory at the Laboratory Centre for Disease Control, Ottawa, Canada, where H serotyping was performed, and specimens were tested for free fecal cytotoxin. If the Premier EHEC test was positive and the E. coli O157:H7 culture was negative, VTEC strains other than E. coli O157:H7 (non-O157 VTEC) were sought in the reference laboratory.
EIAs.
These were performed in the site laboratories and in the reference laboratory, according to the manufacturer’s instructions. Briefly, the Premier E. coli O157 EIA detects the presence of E. coli O157 lipopolysaccharide. A stool extract is placed in a well coated with anti-O157 antibody, to capture the antigen. After a 20-min incubation period, the well is washed and a peroxidase-conjugated antibody is added. A second 10-min incubation is followed by the addition of substrate for color development, measured spectrophotometrically. The threshold optical density reading for a positive result is 0.3 U.
The Premier EHEC EIA detects the presence of VT1 and VT2. For maximum sensitivity, the stools are incubated in MacConkey broth at 37°C for 16 to 24 h before testing. The broth supernatant is added to the antibody-coated wells. After a 1-h incubation period, detection antibody, enzyme conjugate, and substrate solution are added at 30-min intervals.
Cytotoxin.
Stool cytotoxin tests were performed as previously described (13).
Non-O157 VTEC detection.
VT-producing clones were detected by the colony dot blot procedure as follows. Stools were cultured in MacConkey broth (Difco, Detroit, Mich.) overnight at 37°C. Tryptic soy agar plates containing 25 ng of mitomycin C (Sigma, St. Louis, Mo.) per ml were prepared. Two 0.34-μm-pore-size nitrocellulose filters (Schleicher & Schuell, Keene, N.H.) were placed on top of each other on the tryptic soy agar plate. The MacConkey growth was plated directly onto the top filter and then incubated overnight at 37°C. After overnight growth, the lower nitrocellulose filter was removed from the plates. The filters were blocked for 1 h in 5% nonfat dry milk (Carnation, Los Angeles, Calif.) in a buffer composed of 10 mM Tris, 150 mM NaCl, and 0.05% Tween 20 (TNT), at pH 8.0 at room temperature. They were then washed once and incubated for 1 h at room temperature with 1 μg of 4D1 monoclonal antibody (gift from Meridian Diagnostics Inc.) per ml and in TNT followed by three washes with TNT. Alkaline phosphatase-labelled goat anti-mouse immunoglobulin G (gift from Meridian Diagnostics Inc.) diluted 1:10,000 in TNT was added and incubated for 1 h at room temperature. After three washes with TNT, the filter was developed in 5-bromo-4-chloro-3-indolylphosphate toluidinium (BCIP)–nitroblue tetrazolium phosphatase substrate (Kirkegaard & Perry Laboratories, Gaithersburg, Md.). The addition of water was used to stop the color development. The colonies on the top filter, corresponding to those producing blue color in the assay, were removed and grown on MacConkey agar overnight at 37°C. Isolates were serotyped, and VT-encoding genes were detected by PCR (13), with whole-cell extracts.
Anti-O157 antibodies.
Sera collected on days 0 and 8 were tested for O157 antibody by a hemagglutination test (2). The lowest final serum dilution was 1/500.
Paired patient sera giving a titer of ≤1:500 in the first serum, taken at presentation, and a titer of ≥1:1,000 in the second serum, taken on the eighth day of illness, were considered to indicate recent infection with E. coli O157 (1, 3, 5). Negative control sera from 104 control subjects aged 6 months to 16 years with no history of recent diarrhea all yielded titers of ≤1:500.
Definition of field test results with Premier E. coli O157 and Premier EHEC.
The results obtained in the site laboratories (field results) were designated according to the following criteria: (a) true positives, EIA positive and E. coli O157:H7 isolated or significant rising titer to O157 polysaccharide antibody shown in paired sera; (b) true negatives, EIA negative and E. coli O157:H7 not isolated; (c) false negatives, EIA negative and E. coli O157:H7 isolated; (d) false positives; Premier E. coli O157 test positive and E. coli O157:H7 not isolated. The Premier EHEC test was considered to have given a false-positive result only after an adequate search for and failure to find non-O157 VTEC.
Resolved results.
Discordant results which were resolved on repeat testing were reassigned as resolved results (14).
RESULTS
Sensitivities and specificities of Premier E. coli O157.
Table 1 shows the sensitivities and specificities of the Premier E. coli O157 test against the gold standard of E. coli O157:H7 isolation or antibody rise (one patient). All values for the field results are ≥86% with the exception of the positive predictive value in the field, which was 81%. This was because there were 13 stools yielding apparent false-positive Premier E. coli O157 results, 10 of which were reassigned for the resolved results (Table 2). Eight of these were negative on repeat testing and were reassigned as true negatives for the resolved results. Figure 1 shows a comparison between the optical density readings given by the true positives and those given by these eight specimens. All were at the lower end of the positive range, in contrast to the true positives, most of which were at the maximum reading of 3.0 U. Table 2 presents the field results which were reassigned as resolved results after further testing.
TABLE 1.
Performance of Premier E. coli O157 test in the field and in resolved results after repeat testinga
| Category | Result [no. of specimens with correct result/ total no. of specimens (%)]
|
|||
|---|---|---|---|---|
| Sensitivity | Specificity | PPV | NPV | |
| Field | 57/66 (86) | 798/811 (98) | 57/70 (81) | 798/807 (99) |
| Resolved | 60/68 (88) | 806/809 (99) | 60/63 (95) | 806/814 (99) |
Sensitivity = (number of true positives/number of true positives + number of false negatives) × 100; specificity = (number of true negatives/number of true negatives + number of false positives) × 100; positive predictive value (PPV) = (number of true positives/number of true positives + number of false positives) × 100; negative predictive value (NPV) = (number of true negatives/number of true negatives + number of false negatives) × 100.
TABLE 2.
Premier E. coli O157 field results reassigned for resolved resultsa
| Field result | Reassign- ment | No. | Reason |
|---|---|---|---|
| FP | TN | 8 | Field false positive; repeat test negative |
| FP | TP | 2 | Field false positive but also Premier EHEC positive |
| FN | TP | 1 | Field false negative; repeat test positive in reference laboratory |
FP, false positive; TN, true negative; TP, true positive; FN, false negative.
FIG. 1.
Distribution of optical density readings given by 52 true-positive and 8 apparent false-positive Premier E. coli O157 EIA specimens.
Premier EHEC results.
The field sensitivity of the Premier EHEC test, with respect to the gold standard of E. coli O157:H7 isolation or antibody rise, was 50 of 56 (89%). On repeat testing, one field false-negative result which was positive in the reference laboratory and two positive specimens which were also positive in the Premier E. coli O157 test were reassigned as true positives for the resolved results, in which the sensitivity was 53 of 58 (91%). There were 757 Premier EHEC true negatives with respect to the gold standard of E. coli O157:H7 isolation.
Non-O157 VTEC was isolated from 10 of 17 specimens which were Premier EHEC positive and negative on culture for E. coli O157:H7. The serotypes of these 10 isolates were O26:H11 (four isolates) and O26:NM, O45:H2, O111:NM, O103:H25, O121:H19, and O165:H25 (one isolate of each). Of the remaining seven specimens, three were not investigated further, two failed to yield any VTEC isolates but were also Premier E. coli O157 positive, and one yielded no growth at all on repeat culture. One was negative on both VTEC culture and Premier E. coli O157 testing and was assigned as a false positive. Campylobacter sp. was isolated from this specimen (see below, “Alternate pathogens”).
Correlation between Premier EHEC EIA and stool cytotoxin.
Stool cytotoxin tests were performed on 57 specimens which were positive on culture for VTEC of any serotype (47 E. coli O157:H7 specimens and 10 other VTEC specimens). Forty-four specimens were cytotoxin positive and Premier EHEC positive. Thirteen specimens were Premier EHEC positive and cytotoxin negative: VTEC was isolated from 12 of these, and 1 was VT-PCR positive. None were cytotoxin positive and Premier EHEC negative.
Alternate pathogens.
Gastrointestinal pathogens other than VTEC were isolated from 180 stools as follows: 89 of Campylobacter sp., 77 of Salmonella sp., 8 of Shigella sp., and 6 of rotavirus. The eight site false-positive Premier E. coli O157 specimens which were reassigned as true negatives after repeat reference tests included five Campylobacter sp. isolates and one Salmonella sp. isolate. One unconfirmed false-positive Premier EHEC result was associated with Salmonella sp., and the only definite false positive was associated with a Campylobacter sp. isolation.
DISCUSSION
The objective of this study was to examine the performance of two EIAs as criteria for recruitment of subjects into a prospective study of a therapeutic intervention to prevent HUS. The field results obtained at the investigation sites show that the sensitivities and specificities of both tests are high.
The problem of discordant results has been addressed by the development of two scenarios. The field results are the unmodified results initially reported from the participating sites and should be regarded as a measurement of the actual performance of the tests under the realistic working conditions of our study. These results were generated 24 h per day, in routine laboratories in eight different institutions of varying size, and read blind, and clinical decisions were made on the basis of the rapid test results. The resolved results involved repeat testing and the reassignment of discordant results to the category favoring the test where this was justified. Where repeat tests were not performed, the original discordant results were retained in the resolved results.
The redesignation of the discordant results does not significantly affect the sensitivity or the specificity of the Premier E. coli O157 test, but the positive predictive value is significantly affected by the redesignation of 10 results which were apparently false positive in the field and reassigned for the resolved results (Tables 1 and 2). Eight of these 10 specimens were positive on site testing, negative on repeat testing, and culture negative for E. coli O157:H7. Campylobacter sp. was isolated from five of these, and Salmonella sp. was isolated from one. Incomplete washing may result in adherence of residual fecal material in the wells which could bind nonspecifically to the added O157 conjugate antibody and give a false-positive reading. The physical characteristics of stools from patients with Salmonella and Shigella infection might predispose to this. All the optical density readings from these eight specimens were at the lower end of the range for being positive (Fig. 1), and this would be compatible with nonspecific binding. There is therefore strong evidence that the Premier E. coli O157 test results on these eight specimens were false positives attributable to incomplete washing and can be regarded as true negatives for the resolved results.
Two specimens were negative on culture for E. coli O157:H7, and for non-O157 VTEC, but were positive on both rapid tests. All four results were designated as false positives for the field results and as true positives for each test for the resolved results.
A calculation of the true sensitivity of Premier EHEC would require that all stools in the study would be examined for non-O157 VTEC and for stool cytotoxicity. The EHEC test is designed to measure the presence of stool verocytotoxin. Therefore, the logical gold standard for evaluation of this test is the demonstration of tissue culture cytotoxicity with specific neutralization. This is an expensive procedure, with limited diagnostic usefulness because it takes 48 h, and it is not in the test menu of most routine microbiology laboratories. In our study, although it was feasible to perform cytotoxicity tests on EHEC-positive specimens, obtaining complete cytotoxicity data to calculate sensitivity would have required approximately 900 cytotoxicity tests. This was not feasible, and the sensitivities, 89% in the field and 93% in the resolved sensitivity, are reported against a gold standard of E. coli O157:H7 isolation or serologic response. It would be expected that the sensitivity of this test for the detection of VTEC organisms would be governed by the amount of toxin production, and our results appear to reflect the consistency of VT production by E. coli O157:H7.
The recovery of non-O157 VTEC from 10 of 17 Premier EHEC-positive, E. coli O157:H7 culture-negative specimens is highly significant. Of the remaining seven positive stools, three were not further tested. Given that the test is designed to detect VT, it is reasonable to exclude these three from analysis. One specimen gave no growth at all on repeat culture. It is possible that the patient had received a high dose of an antibiotic which might have prevented organism growth and toxin synthesis in the repeat test. Two specimens were positive in the Premier E. coli O157 test. Only one specimen was positive with the Premier EHEC test and was negative for all VTEC strains, and this was the only definite false-positive specimen. Although a formal expression for the specificity of Premier EHEC cannot be derived because of an unavoidable inconsistency in the gold standard, the findings of 757 true negatives with respect to E. coli O157:H7 isolation, 50 field result true positives, and only 1 definite false positive, together with the cytotoxicity test results (below), are all evidence for the high specificity of this test.
VTEC was isolated from 12 of 13 specimens which were Premier EHEC positive and cytotoxin negative and 1 was VT-PCR positive but culture negative. The EIAs were performed on fresh specimens, but the cytotoxin tests were performed in the central laboratory after freezing, transporting, and thawing. VT may have been present in some of the stool specimens at the time the Premier EHEC test was performed but absent when the tests were performed in the reference laboratory. The absence of cytotoxin-positive, Premier EHEC-negative results, however, supports the manufacturer’s claim for the sensitivity of the test.
In most routine laboratories, Salmonella sp. and Campylobacter sp. would outnumber VTEC strains in specimens tested except in outbreaks. The question of specificity with respect to these pathogens is therefore important. Apart from the discordant results in the Premier E. coli O157 test, which are attributed to incomplete washing and resolved by repeat testing, the incidence of false positives for patients with Salmonella, Campylobacter, and Shigella spp. is no higher than that in specimens negative for these pathogens.
The EIA results are presented here in isolation, but the tests were performed in order to maximize efficiency of patient recruitment into a therapeutic trial one important feature of which was the early administration of study medication, without delay in waiting for the routine stool culture results. In an earlier phase of the evaluation of SYNSORB Pk in which we did not use the rapid tests, 211 of 347 (61%) patients had to be excluded because their stools were culture negative for E. coli O157:H7. The data presented in this paper were generated in year 1 of phase 3 of the trial. The use of the two rapid tests was associated with a greatly improved efficiency in patient recruitment. High sensitivity and negative predictive values of these tests have ensured that few eligible patients were not recruited, and the high specificity and positive predictive values have minimized invalid recruitment of patients who later had to be excluded.
Our results confirm and extend the findings of Kehl et al. (7), who reported that the Premier EHEC test was positive in 100% of six stools from which E. coli O157:H7 was isolated; however, this study did not include the Premier E. coli O157 test. In the same paper, three non-O157 VTEC strains were found and three Premier EHEC test results were reported as false positives. The false positives, however, were given by stools with neutralizable cytotoxin present, which suggests the possibility that VT-producing organisms were present but not detected. In a second study, the Premier EHEC test found VTEC in 11 of 270 successive specimens, of which 6 were E. coli O157:H7 and 5 were non-O157 VTEC strains which would have been missed without this test (11).
Traditional culture methods, including sorbitol-MacConkey agar, do not reveal the presence of non-O157 VTEC, and the techniques needed for the detection of these organisms are beyond the scope of most routine laboratories. At this time, therefore, rapid immunological tests are the most economical way to address this problem, and this study demonstrates the practicability of using the Premier EHEC test for this purpose in a routine laboratory situation. VT production by members of the family Enterobacteriaceae other than E. coli has been described elsewhere (12), and the availability of a simple test for the presence of VT will determine the true prevalence of non-O157 VTEC and of toxin production by organisms other than E. coli.
ACKNOWLEDGMENTS
We acknowledge the following SYNSORB Pk Study coinvestigators: Warren Anquist, Matsqui-Sumas-Abbotsford General Hospital, Abbotsford, British Columbia, Canada; Verna Yiu and Frances Harley, University of Alberta Hospital, Edmonton, Alberta, Canada; Cheri Nijssen-Jordan and Julian Midgley, Alberta Children’s Hospital, Calgary, Alberta, Canada; Malcolm Ogborn, Health Sciences Centre, Winnipeg, Manitoba, Canada; Paul Goodyer, Montreal Children’s Hospital, Montreal, Quebec, Canada; and Jean Turgeon and Francois Proulx, L’Hôpital Sainte-Justine, Montreal, Quebec, Canada. We thank Elaine Paice for her administrative assistance.
This study was supported by Meridian Diagnostics Inc. and SYNSORB Biotech Inc., Calgary, Alberta, Canada.
REFERENCES
- 1.Barrett T J, Green J H, Griffin P M, Pavia A T, Ostroff S M, Wachsmuth I K. Enzyme-linked immunosorbent assays for detecting antibodies to shiga-like toxin I, shiga-like toxin II, and Escherichia coli O157:H7 lipopolysaccharide in human serum. Curr Microbiol. 1991;23:189–195. [Google Scholar]
- 2.Bitzan M, Karch H. Indirect hemagglutination assay for diagnosis of Escherichia coli O157 infection in patients with hemolytic-uremic syndrome. J Clin Microbiol. 1992;30:1174–1178. doi: 10.1128/jcm.30.5.1174-1178.1992. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Bitzan M, Moebius E, Ludwig K, Muller-Wiefel D E, Heesemann J, Karch H. High incidence of serum antibodies to Escherichia coli O157 lipopolysaccharide in children with hemolytic-uremic syndrome. J Pediatr. 1991;119:380–385. doi: 10.1016/s0022-3476(05)82049-9. [DOI] [PubMed] [Google Scholar]
- 4.Brotman M, Giannella R A, Alm P F, Bauman H, Bennett A R, Black R E, Bruhn C M, Cohen M B, Gorbach S L, Kaper J B, Roberts M R, Staneck J L, Taylor S, Troutt H F, Bell B P, Buchanan R L, Durham K, Feng P, Forman C T, Galler R G, Gravani R B, Hall R B, Hancock D D, Hollingsworth J, Karmali M A, et al. Consensus conference statement—Escherichia coli O157-H7 infections—an emerging national health crisis, July 11–13, 1994. Gastroenterology. 1995;108:1923–1934. [PubMed] [Google Scholar]
- 5.Chart H, Scotland S M, Rowe B. Serum antibodies to Escherichia coli serotype O157:H7 in patients with hemolytic uremic syndrome. J Clin Microbiol. 1989;27:285–290. doi: 10.1128/jcm.27.2.285-290.1989. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Karmali M A, Steele B T, Petric M, Lim C. Sporadic cases of haemolytic-uraemic syndrome associated with faecal cytotoxin and cytotoxin-producing Escherichia coli in stools. Lancet. 1983;i:619–620. doi: 10.1016/s0140-6736(83)91795-6. [DOI] [PubMed] [Google Scholar]
- 7.Kehl K S, Havens P, Behnke C E, Acheson D W K. Evaluation of the Premier EHEC assay for detection of Shiga toxin-producing Escherichia coli. J Clin Microbiol. 1997;35:2051–2054. doi: 10.1128/jcm.35.8.2051-2054.1997. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Louise C B, Obrig T G. Human renal microvascular endothelial cells as a potential target in the development of the hemolytic uremic syndrome as related to fibrinolysis factor expression, in vitro. Microvasc Res. 1994;47:377–387. doi: 10.1006/mvre.1994.1030. [DOI] [PubMed] [Google Scholar]
- 9.Louise C B, Obrig T G. Specific interaction of Escherichia coli O157-H7-derived shiga-like toxin II with human renal endothelial cells. J Infect Dis. 1995;172:1397–1401. doi: 10.1093/infdis/172.5.1397. [DOI] [PubMed] [Google Scholar]
- 10.Obrig T G, Del Vecchio P J, Brown E, Moran T P, Rowland B M, Judge T K, Rothman S W. Direct cytotoxic action of Shiga toxin on human vascular endothelial cells. Infect Immun. 1988;56:2373–2378. doi: 10.1128/iai.56.9.2373-2378.1988. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Park C H, Gates K M, Vandel N M, Hixon D L. Isolation of Shiga-like toxin producing Escherichia coli (O157 and non-O157) in a community hospital. Diagn Microbiol Infect Dis. 1996;26:69–72. doi: 10.1016/s0732-8893(96)00180-0. [DOI] [PubMed] [Google Scholar]
- 12.Paton A W, Paton J C. Enterobacter cloacae producing a Shiga-like toxin II-related cytotoxin associated with a case of hemolytic-uremic syndrome. J Clin Microbiol. 1996;34:463–465. doi: 10.1128/jcm.34.2.463-465.1996. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Pollard D R, Johnson W M, Lior H, Tyler D, Rozee K R. Rapid and specific detection of verotoxin genes in Escherichia coli by the polymerase chain reaction. J Clin Microbiol. 1990;28:540–545. doi: 10.1128/jcm.28.3.540-545.1990. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Rotbart H A, Ahmed A, Hickey S, Dagan R, McCracken G H, Jr, Whitley R J, Modlin J F, Cascino M, O’Connell J F, Menegus M A, Blum D. Diagnosis of enterovirus infection by polymerase chain reaction of multiple specimen types. Pediatr Infect Dis J. 1997;16:409–411. doi: 10.1097/00006454-199704000-00014. [DOI] [PubMed] [Google Scholar]