1292

Effective Use of the Clinical Microbiology Laboratory for Diagnosing Diarrheal Diseases

Janet Hines and Irving Nachamkin

Diarrheal diseases are an enormous problem worldwide and cause considerable morbidity and mortality, especially in devel­oping nations. Four times as many diarrhea-related deaths occur worldwide in a single day than occur in an entire year in the United States [1, 2]. Individuals who are at greater risk for developing diarrheal diseases than other segments of the popu­lation, especially in developed countries, include those in day­care centers, travelers, and patients with HIV infection. During the past decade, the HIV epidemic has resulted in a relatively large number of persons at high risk for developing infectious diarrhea, particularly that due to organisms heretofore unrecog­nized or unheard of in immunocompetent patients.

As the number of potential etiologic agents of diarrheal dis­ease increases, the clinical microbiology laboratory takes on an increasingly important role in identifying these agents. Of the many tests performed in the clinical microbiology labora­tory, there is virtually no more labor-intensive and costly rou­tine test than that ofthe stool culture and parasitological exami­nation. The microbiological workup of stool requires a specific algorithm to isolate and identify pathogens. Factors including the number of special media required, the labor intensiveness of the process, and the generally low sensitivity of stool cultures combine to make this test one of the most expensive diagnostic tests offered; the cost ranges from $100 to more than $100,000 per positive result, depending on the tests performed, the preva­lence of disease, and number of samples tested [3 - 5].

While stool cultures and parasitological examinations are invaluable tools for epidemiological studies, the role of these cultures in the management of acute diarrhea has not been well defined because many episodes of acute diarrhea are self limited. It is estimated that >4 million tests for bacteria and parasites are performed in the United States annually; thus, efficient use of the laboratory in an era of cost containment is critical [6]. The purpose of this paper is to review the current understanding and limitations of the clinical microbiology labo­ratory in the management of diarrheal illness and to dispel

Received 27 August 1996. J. H. is the William PepperlBecton Dickinson Fellow in Clinical Microbiol­

ogy. Reprints or correspondence: Dr. Irving Nachamkin, University ofPennsylva­

nia, Department of Pathology and Laboratory Medicine, 7th Floor, Founders Building, 3400 Spruce Street, Philadelphia, Pennsylvania 19104-4283.

Clinical Infectious Diseases 1996; 23:1292-301 © 1996 by The University of Chicago. All rights reserved. 1058--4838/96/2306-0032$02.00

From the Departments of Pathology and Laboratory Medicine (Clinical Microbiology) and Medicine (Infectious Diseases), University of

Pennsylvania School of Medicine, Philadelphia, Pennsylvania

some common beliefs about the usefulness-or lack thereof­of certain commonly performed tests.

The Role of Empirical Treatment Without Laboratory Support

Although there are epidemiological and public health con­cerns about tracking enteric pathogens and identifying potential outbreaks of infection, empirical therapy in the absence of microbiological evaluation is warranted for certain patient pop­ulations with acute enteric infections. The management oftrav­eler's diarrhea is a good example [7]. Furthermore, several studies have shown that empirical therapy for acute diarrheal disease (particularly with fiuoroquinolones), regardless of the infecting organism, reduces number of stools as well as other signs and symptoms [8-11].

The downside of empirical therapy is that treatment can prolong carriage of Salmonella species and foster recurrence of the infection [12 -14]. In addition, increasing resistance to fiuoroquinolones, particularly among organisms such as Campylobacter species, may reduce the efficacy of empirical treatment, necessitating a more important role for the laboratory in directing specific therapy based on susceptibility results [15]. Studies are needed to examine the cost-benefit of empirical therapy, with or without laboratory support, to better define those patients who would benefit from having specific microbi­ological tests performed.

For Which Patients Should Routine Stool Cultures and Ova and Parasite Examinations Be Performed?

It has been common practice to request stool cultures and! or parasitological examinations for any patient with a diarrheal illness. However, in recent years it has become increasingly apparent that such routine testing is clinically ineffective and cost ineffective. Organisms such as Salmonella, Campylo­bacter, and Shigella, which are the most common causes of sporadic diarrheal disease in the United States, are for the most part community acquired and rarely cause nosocomial infections. The same holds true for parasites. Clostridium diffi­cile, on the other hand, causes both community-acquired and nosocomial diarrheal disease. Routine stool cultures are fre­quently performed for hospitalized patients despite the fact that the yield of such cultures is low [3, 6, 16-20].

em 1996;23 (December) Microbiological Diagnosis of Diarrhea 1293

Table 1. Summary of results of stool analyses for outpatients and inpatients, September 1987 through August 1988, at the University of Pennsylvania Medical Center (reprinted with permission from the Journal of the American Medical Association [6].

Inpatients hospitalized 0-3 Inpatients hospitalized > 3 Outpatients days days

No. positive No. positive No. positive All patients No. tested (Percent total, No. tested (Percent total, No. tested (percent total, (Percent percent (Percent percent (Percent percent No. positive

Type of test workload)* yield)t workload)* yield)t workload)* yield)t No. tested (Percent yield)

Bacterial culture 569 (29) 32 (80, 6) 398 (20) 8 (20, 2) 997 (51) 0(0,0) 1,964 40 (2) Ova/parasite 622 (44) 33 (92, 5) 269 (19) 3 (8, I) 532 (37) o (0, 0) 1,423 36 (3) C. difficile* 84 (3) 28 (5, 33) 516 (19) 88 (16, 17) 2,068 (78) 444 (79, 21) 2,668 560 (21)

* Percent workload = no. of specimens tested for patient group/total no. of specimens submitted for all patient groups. t Percent total = no. of positive results/total no. of positive results for all patient groups; percent yield = no. of positive specimens/no. of specimens submitted

for that patient group. * Data for C. difficile assay were collected during a 3-month period and extrapolated to 1 year for comparison.

On the basis of studies performed at our institution [6], we recommended that routine stool cultures and parasitological examinations be limited to outpatients or inpatients who have been hospitalized for ~3 days; we call this recommendation the "3-day rule" (table 1). We found that routine stool cultures that are performed for patients hospitalized for > 3 days ac­counted for >50% of the laboratory's workload, and these cultures failed to yield a single previously unrecognized organ­ism over a 3-year period. Other studies have confirmed this recommendation [16-18, 20], and the College of American Pathologists (CAP) recently recommended that the 3-day rule be used for routine stool cultures and that a 4-day rule be used for parasitological examinations [3]. The use of such recom­mendations may result in savings estimated as ranging from $27 million to $73 million annually [6, 20].

What Is a Routine Stool Culture?

In an attempt to control costs and time, specimen workups in microbiology laboratories are targeted at only those pathogens important in the respective patient population. Laboratorians and clinicians should work together to establish the common enteric pathogens present in their patient population and use that information to guide routine workups.

Campylobacter, Salmonella, and Shigella are the most com­mon bacterial agents of diarrheal disease and should always be sought in routine stool cultures. A recent CAP survey [3] showed that most laboratories (99.3%) included Salmonella and Shigella in the routine stool workup, but only 96% rou­tinely included Campylobacter. Thus, there are still some labo­ratories that do not include even the most common pathogens in the routine workup. Other organisms such as Aeromonas, Plesiomonas, Yersinia, Escherichia coli 0157, and Vibrio were included in the routine culture in 30%-60% of the laboratories in the survey. For example, our laboratory does not routinely

examine all stool samples for Vibrio parahaemolyticus or Yer­sinia enterocolitica because of the results of previous surveys of our patient population. However, if a patient presented with a syndrome consistent with appendicitis due to Yersinia or had visited an area where V. parahaemolyticus is seen more commonly (i.e., the Chesapeake Bay area), notification of the laboratory would result in the proper evaluation for these organ­isms.

In a recent survey of all stool specimens submitted to our laboratory during a I-year period for detection of E. coli 0157, only one sample was positive (authors' unpublished data). A survey of clinical laboratories by the Association of State and Territorial Public Health Laboratory Directors found that only 29% of surveyed laboratories screened all stool samples for E. coli 0157:H7 and that 43% screened selected samples (e.g., bloody stools or in cases where the physician has made a specific request) [21]. However, E. coli 0157 was reported to be the most frequently isolated organism at the University of Calgary Clinical Microbiology Laboratory (Calgary, Alberta, Canada), accounting for 34% of all positive stool cultures [18].

Microbiology laboratories do not have uniform protocols for detecting enteric pathogens. Several types of differential and selective media are used in the workup of stool samples when a particular etiologic agent is sought [3]. In general, multiple culture media are used, since there is no single medium avail­able that is 100% sensitive. Thus, the decision to use certain culture media is usually based on both published data and personal preference [22].

Diagnostic Strategies for the Treatment of Patients with Diarrhea

Few studies have examined the cost and clinical effective­ness of different diagnostic strategies for the workup of diar-

1294 Hines and Nachamkin eID 1996;23 (December)

rheal diseases, although numerous articles have been published on the clinical management of diarrheal illnesses [23].

In one of the few published studies in which diagnostic strategies were addressed, three evaluations for the manage­ment of diarrheal disease in patients with AIDS were compared [24]. When treatment strategies were compared on the basis of the results of a full evaluation (stool culture for enteric pathogens, three ova and parasite examinations, two blood cul­tures, and endoscopy with biopsy), a limited evaluation (stool culture, three ova and parasite examinations, and two blood cultures), and a minimal evaluation (stool culture only), the clinical response rate among the patients was similar, regardless of the diagnostic strategy used. However, the overall cost dif­fered markedly between the strategies used, and it is not surpris­ing that the full evaluation was more than three times as expen­sive as the minimal and limited evaluation strategies.

Several clinical algorithms have recently been suggested for predicting C. difficile disease in hospitalized patients [25, 26]. Manabe et al. [25] identified significant variables such as the presence of fecallactoferrin, a positive gram stain, use of ceph­alosporins, and the passing of semiformed stools in their model, but the ability of these variables to predict positivity for C. difficile were not assessed. Cooper et al. [26] found that periph­eralleukocytosis, the presence of leukocytes in stool, and ante­cedent hospitalization were significant variables in their model and that these variables showed a good correlation between the number of predictors present and frequency of positive toxin results. Fecal leukocytes were present in only 24% of patients whose stool samples were positive for C. difficile toxin, which severely limited the usefulness of this predictor.

Assessment of Testing Strategies for Cost-Effective Laboratory Workup of Stool Samples

Given the forces shaping medical economics in the 1990s, various strategies have been proposed to decrease laboratory costs [5, 6, 17,22,27,28]. These strategies have been based on the following factors: (1) the process of a stepwise or tiered workup based on the most likely to least likely pathogens; (2) categorization of the clinical presentation as inflammatory or noninflammatory to distinguish between patients who do and do not need cultures performed; (3) the identification of certain clinical and epidemiological circumstances under which a par­ticular group of pathogens may be excluded; and (4) the degree to which identification of a pathogen will affect the treatment of the patient. Despite the existence of these strategies, there is no clear consensus on a uniform algorithm.

The use of a least likely to most likely strategy for workup of stool samples is a reasonable approach to the diagnostic workup of a patient with diarrheal disease. As mentioned above, common etiologies should be sought first, and when the results are negative, less common agents should be pursued, especially if the illness has not spontaneously resolved. Mild

diarrhea tends to dictate its own workup and, in general, when the diarrhea stops, so do the specimens. For routine purposes, the laboratory should have a policy of rejecting samples of formed stool, since even a positive result will have a minimal impact on cases of resolving diarrhea. However, there should be exceptions to such a policy; e.g., follow-up cultures should be performed for Salmonella carriers because there is a need to demonstrate clearance of the organism. A recent CAP survey of laboratories demonstrated that fewer than 10% had any pol­icy for limiting the number or types of samples processed for routine culture [3]. The cost-effectiveness or clinical effective­ness of a stepwise model of testing for the presence of bacterial or parasitic agents, although seemingly intuitive, has not yet been determined.

The practice of dividing infections into inflammatory vs. noninflammatory presentations is based on the supposition that a patient with bloody stools, tenesmus, fever, persistent abdom­inal tenderness, and the presence offecalleukocytes is infected with an invasive bacterial pathogen and will require antimicro­bial therapy [29]. This classification is limited by the fact that there is significant overlap among the syndromes caused by bacterial pathogens: some invasive bacteria cause watery diar­rhea without fever, blood, or abdominal pain, as would be typical in cases of noninflammatory diarrhea. This distinction is probably most important in designating those patients most likely to benefit from treatment; however, to our knowledge, the clinical benefit of this differentiation has never been proven.

One obvious approach to limiting the workup for enteric pathogens is to consider clinical and epidemiological factors that would exclude certain pathogens. For example, requesting cultures for Vibrio species in the absence of a specific exposure (i.e., ingestion of shellfish or travel to areas of endemicity) is an excess that can no longer be afforded in the initial workup of stool samples. Requests for cultures to detect Staphylococcus aureus and yeasts that are unlikely causes of diarrhea should be limited as well.

Strategies for determining the cost-effectiveness of identi­fying specific pathogens and for determining how this identifi­cation affects clinical outcome have not been published, al­though physicians make this decision intuitively. Empirical therapy for traveler's diarrhea is again an example of a situation in which specific identification of a pathogen will have little impact on outcome.

Is There a Role for Rapid, Nonculture Screening Tests?

Examination of stool for the presence of leukocytes began in the early twentieth century when it was common practice to examine stool for protozoa, but the clinical significance of detecting leukocytes has only recently been reevaluated. Clinicians have been routinely taught that the presence of fecal leukocytes is an indicator of "inflammatory" diarrhea and

em 1996;23 (December) Microbiological Diagnosis of Diarrhea 1295

Table 2. Detection of fecal leukocytes during infection with com­mon bacterial enteric pathogens.

Mean percentage

No. of positive References Pathogen patients (range)

[8,27,32] Campylobacter species 194 58 (25-80) [32,33] Escherichia coli 112 54 (42-65)

0157:H7 [8, 17, 32, 34] Salmonella species 140 52 (11-82) [8, 17, 27, 32, 34, 35] Shigella species 252 73 (49-100) [32] Yersinia enterocolitica 27 48 [25, 26, 36, 37] Clostridium difficile 160 42 (24-63)

should be used to determine whether culture of stool samples is needed [5, 29].

The algorithm proposed by Guerrant is the best example of the role of this finding [29]. In this algorithm, analysis offecal leukocytes, and, more recently, fecal lactoferrin [30], figure prominently in the approach to the diagnosis and management of infectious diarrhea. Only those samples containing leuko­cytes are cultured, and only in such cases is the need for antimi­crobial therapy evaluated. Donowitz et al. [31] also recommend fecal leukocyte analysis in the initial evaluation of patients with chronic diarrhea, although the rationale for this recom­mendation is not obvious. Although it is commonly believed that the fecal leukocyte test is useful for identifying patients with bacterial gastroenteritis caused by the most common agents of putatively inflammatory diarrhea (i.e., species of Campylobacter, Salmonella, and Shigella), the data do not nel:­essarily support a role for this test.

Over the years, a number of studies have been performed that contain data on fecal leukocyte determinations [8, 17,25-27, 32-37]; the data are summarized in table 2. Even for detection of the prototypical inflammatory diarrheal disease agent, Shigella, the sensitivity of the fecal leukocyte test is on average 73%. For the other agents, the sensitivity is ~50%. In one recent study, the positive and negative predictive values of the fecal leukocyte test were at best 45% and 93%, respec­tively [38]. While studies have shown that the yield of positive cultures increases for patients in whom fecal leukocytes are detected [27], the performance characteristics of this test are not sufficient for its use as a screening procedure.

Although the cost of a fecal leukocyte examination is low, the clinical effectiveness ofthe test is not known. To our knowl­edge, studies examining the impact of fecal leukocyte tests on clinical outcome have yet to be done. If one were to find that patients with fecal leukocytes were more likely to benefit from antimicrobial therapy than those without fecal leukocytes, a case might be made for doing the test.

Another test that has received some attention as an alterna­tive to the fecal leukocyte test is the fecal lactoferrin assay.

Lactoferrin is an iron-binding glycoprotein found in secondary granules in leukocytes, and it is detected with use of a commer­cially available latex agglutination test [30]. In a recent small study conducted by Choi et al. [30], the sensitivity of the lactoferrin assay in detecting species of Salmonella, Campylo­bacter, and Shigella varied from 83% to 93%, and its specificity varied from 61 % to 100%, depending on the cutoff used for positivity. Silletti et al. [39] found that compared with culture, the fecal lactoferrin assay was more sensitive but less specific than the fecal leukocyte test: the fecal lactoferrin was 85% sensitive and 79% specific, while the fecal leukocyte test was 31 % sensitive and 96% specific.

Fecal leukocyte and lactoferrin assays have also been studied as screening tests for C. difficile. Marx et al. [37] found that the sensitivity of fecal leukocyte analysis was only 28% among patients positive for C. difficile toxin. Yong et al. [36] found that fecal lactoferrin was more sensitive as a marker for C. difficile toxin-positive patients than was fecal leukocyte analysis (75% vs. 40%, respectively), but fecal leukocyte analy­sis was more specific (92% vs 46%, respectively).

In a recent study by Manabe et al. [25], both fecal leukocyte and lactoferrin assays had poor performance characteristics as markers for C. difficile toxin in stool samples with positive predictive values and negative predictive values ranging from 25% to 27% and from 89% to 90%, respectively. Cooper et al. [26] found that while fecal leukocytes were more often found in patients positive for C. difficile toxin, only 24% of toxin-positive patients had fecal leukocytes detected. Thus, even for C. difficile disease, these tests have little proven diag­nostic efficacy, and we do not recommended performing them.

While the rapid tests described above are of limited proven usefulness, the gram stain still has a role in the diagnosis of diarrheal disease. Campylobacter jejuni, one of the most com­mon causes of bacterial gastroenteritis, can be easily visualized in a stool smear, especially if the counterstain, safranin 0, is replaced with basic fuchsin. The sensitivity of the gram stain has been reported as ranging from 66% to 94%, with high specificity [40]. Unfortunately, the gram stain has no value for detecting any other common enteric pathogen, since these can­not be differentiated from normal stool flora with use of the gram stain.

General Processing and Sampling Considerations

A few studies have been conducted to determine the number of specimens needed for optimal detection of enteropathogens. For detecting common enteropathogens, one or two samples appear to be adequate for episodes of acute disease. Church et al. [18] found that for children, 97% of enteropathogens were detected with one fecal sample. In a recent CAP survey of 601 institutions where both stool cultures and parasite examinations were performed, it was determined that testing of two stool specimens resulted in detection of99% of bacteriologic agents,

1296 Hines and N acharnkin em 1996;23 (December)

and testing of three specimens resulted in detection of 99.8% of parasites [3]. In another study by Marti and Koella [41], the false-negative rate for three ova and parasite examinations varied from 0.1 % (Trichuris or Ascaris species) to 5.8% (Enta­moeba histolytica). Approximately 91 % of parasites were de­tected using a single sample in a study from Duke Medical Center [16].

Valenstein and colleagues [3] recently recommended that no more than two stool specimens per patient for bacteriology and two or three samples per patient for parasitology be the standard for routine workups and that additional testing be performed only after consultation with the laboratory. The fact that the cost of detecting additional cases of disease with use of more samples is several times higher than that with a single sample (i.e., $117 for one sample vs. $116,636 for four samples) sup­ports this recommendation.

For detection of C. difficile disease on the basis of the pres­ence oftoxin, two stool samples were shown to be more sensi­tive than was a single sample [25]. On the basis of our own in-house data from toxin B cytotoxicity cell culture assays, we recommend that up to three samples be tested to rule out disease (Nachamkin et aI., unpublished observations).

In a recent study by Renshaw et al. [42], the sensitivity of a single assay for C. difficile toxin was 94%, with as many as four samples needed for 100% sensitivity (based on all positive tests). Repeated assays for C. difficile toxin within a 7-day period accounted for one-third of all testing per­formed at these authors' institution, but such assays provided clinically useful information in only 1 % of cases. On the basis of this information, it would seem reasonable to send two samples to the laboratory to rule out C. difficile infection, but such samples should be spaced several days apart before testing. Additional testing (more than two specimens) for C. difficile infection should be considered only after consulta­tion with the laboratory.

Stool samples are the most common specimens submitted to the laboratory for bacteriologic and parasitological examina­tions. Freshly obtained specimens that are processed soon after collection (i.e., < I hour) are optimal for processing; however, this procedure is not practical in most situations. Therefore, all specimens should be sent to the laboratory in transport devices specifically designed for bacteriologic and parasitological ex­aminations. Of the many transport media available, Cary-Blair medium appears to be the best medium overall for supporting the isolation of the most commonly encountered bacterial agents [40]. Recommendations for cost-effective testing for enteric pathogens are shown in table 3.

Recent Advances

Bacterial Agents

Because the variety of bacterial agents that cause diarrheal diseases is wide, little progress has been made in replacing

conventional culture methods in the clinical microbiology labo­ratory. The most promising approaches appear to involve mo­lecular techniques for detecting specific organisms or their tox­ins, but none have been commercialized or are available for routine testing [43]. A brief summary of commonly considered enteropathogens is given below.

Aeromonas species and Plesiomonas shigelloides. The role of these organisms as enteropathogens has never been com­pletely resolved, and the evidence for their pathogenicity is based on clinical, epidemiological, and in vitro studies. The best clinical correlates are observed when samples contain moderate-to-heavy growth of the organisms in the absence of other known enteropathogens. However, human challenge stud­ies with these organisms have not produced disease. Three species of Aeromonas are considered the major pathogenic species and include A. hydrophila, A. caviae, and A. veronii biovar sobria [44]. These species can be isolated on nonselec­tive media (i.e., sheep blood agar) if these media are used in a routine culture setup; thus, the need for special selective media is uncertain [22].

Campylobacter and Arcobacter species. Many advances have been made over the past 5 -10 years in diagnosing campy­lobacter infections and describing new species that cause diar­rheal diseases. Two genera, Campylobacter and Arcobacter, are included in the recently adopted family Campylobacteraceae. There are now 19 species and subspecies within the genus Campylobacter and four species in the genus Arcobacter, most of which have been described as causing or being associated with enteric infection [40]. Some species of Campylobacter other than C. jejuni and C. coli that may be of particular impor­tance include C. fetus subspecies fetus, c. upsaliensis, and Arcobacter butzleri. Helicobacter cinaedi and Helicobacter fennelliae, previously designated within the genus Campylo­bacter, are also important causes of proctitis and proctocolitis in immunocompromised patients.

Of particular note is the emergence of C. jejuni infection as a cause of Guillain-Barre syndrome [45]. As mentioned elsewhere, routine stool cultures should always include the isolation of Campylobacter jejuni and C. coli, since these are the most commonly recognized Campylobacter species that cause diarrheal disease. A variety of culture methods such as filtration on nonselective media are needed to isolate the other species of Campylobacter and should be considered for patients with persistent diarrheal disease from whom no other identifi­able agent has been recovered.

C. difficile. In contrast to the other enteropathogens, C. diffi­cile is the most commonly recognized bacterial agent of noso­comial diarrheal disease. Several approaches to the laboratory diagnosis of C. difficile disease have been published, and tests include detection of C. difficile enterotoxin (toxin A or the cytotoxin, toxin B). Additional methods include culture, detec­tion of other cellular antigens, and, more recently, molecular methods [46, 47]. At the present time, molecular methods are

CID 1996;23 (December) Microbiological Diagnosis of Diarrhea 1297

Table 3. Recommendations for cost-effective testing for enteric pathogens.

• Routine cultures should always include Campylobacter,

Salmonella, and Shigella; other agents included in the routine culture panel should be detennined on the basis of in-house or regional survey data.

• Submit additional samples for less common bacteriologic agents as a second-line test after the common agents have been ruled out.

• Up to two stool samples should be submitted for bacteriologic culture, and three samples should be submitted for ova and parasite examination for optimal sensitivity. Wait for the results of the first culture and/or examination before sending additional samples, rather than ordering multiple tests.

• Two to three samples may also be needed to rule out Clostridium difficile disease, and samples should be spaced several days apart.

• Use the 3-day rule for routine stool cultures and/or parasitological examinations for hospitalized patients.

• Because there are new tests for Giardia lamblia, Entamoeba

histoiytica, and Cryptosporidium, consider testing specifically for these agents initially and forego the initial routine ova and parasite examination, especially when other parasitic agents are unlikely.

still used only in the research laboratory and are not available from commercial sources. The choice to use the more common methods will be determined by issues of staffing, cost, and turnaround time (table 4).

In general, detection of C. difJicile toxin B with use of a cell culture-toxin neutralization cytotoxicity assay (TCA) has been considered to be the "surrogate" criterion standard for evaluat­ing the different test methods. It has been difficult to assess diagnostic test performance because a true criterion standard for C. difJicile disease requires both clinical and laboratory criteria, which are frequently not used in most studies [46]. Commercial systems for assaying toxin B are available and appear to be essentially equivalent to those for which in-house reagents are used [48]. The format of the assays is such that experience in cell-culture techniques is not necessary, and these assays can be performed easily in most clinical microbiology laboratories.

A number of EIAs that detect C. difJicile toxin A and/or toxin B are commercially available and have good performance characteristics when compared with TCA, but they are clearly less sensitive than TCA. The sensitivity and specificity ofthese assays have varied considerably, depending on the assay used [46, 49-51], and in some cases, positive results may need to be verified by TCA. EIAs are generally easy to perfonn, and the turnaround time for obtaining results is short.

Some laboratories use culture for detecting C. difJicile [46]. Although culture may be a sensitive method for detecting the organism in stool samples, isolates must be tested for produc­tion of toxin, thus delaying results for several days. Therefore, we do not recommend culture as a routine method for diagnos­ing C. difJicile disease.

Latex agglutination assays that detect a C. difJicile antigen (not toxin) are commercially available, easy to perform, and

rapid; however, they are not as sensitive or specific as TCA or EIA [46, 49, 51]. In addition, a method that detects toxin A or toxin B should be used for patients who are negative for antigen but clinically suspected of having C. difJicile disease.

E. coli. Numerous types of E. coli that cause diarrheal disease have been described, including enterotoxigenic strains, entero­pathogenic strains, enteroinvasive strains, enterohemorrhagic strains, and enteroaggregative strains [52]. Of these different types, only enterohemorrhagic strains of serotype 0157:H7 can be routinely detected in most clinical microbiology laboratories because a specific selective medium, sorbitol-MacConkey agar, is widely available. The geographic distribution of these strains varies, and media for detection may be available in some labo­ratories and not in others. Sorbitol-negative strains can be fur­ther identified with specific serotyping reagents or toxin as­says [22].

The timing of cultures appears to be important: the rate of recovery of E. coli 0157 is highest during the first few days after the onset of diarrhea and drops off significantly after 7 days [33]. Because serotypes other than 0157 can cause dis­ease, diagnostic alternatives such as detection of fecal vero­toxin, recently proposed as Stxl [53], may be necessary for optimal sensitivity [54, 55]. Commercial systems for detecting 0157 directly in stool samples have also been recently de­scribed [56].

Y enterocolitica. Detection of Y enterocolitica continues to be based on conventional culture methods. The need to test for the presence of this organism in routine cultures will depend

Table 4. Characteristics of assays used for detecting Clostridium difficile.

Method

Cytotoxicity assay (toxin B)

ErA (toxin AlB)

Latex agglutination (nontoxin antigens)

Culture

Advantages

Sensitive and specific; bioassay; kits available

Rapid; many kits available

Rapid

Sensitive

Disadvantages

Not as rapid as other methods; may require tissue culture if kits not used

Slightly less sensitive and specific than cytotoxicity assay; positive results may need verification

Does not detect toxin A or toxin B; nonspecific for toxigenic strains; least sensitive and specific; positive results should be confinned by additional tests

Time-consuming; additional tests to distinguish toxigenic from nontoxigenic strains required

1298 Hines and Nachamkin em 1996;23 (December)

on the geographic prevalence of Y. enterocolitica disease. Not all strains of Y. enterocolitica cause diarrheal disease; certain serotypes have been identified as pathogenic (including sero­types 04, 32; 08, 013a, 13b; 018, 020; and 021) and are common in the United States. Virulence is associated with the presence of a 40- to 48-megadalton plasmid. A number of phenotypic tests including the presence of pyrazinamidase, sali­cin fermentation, esculin hydrolysis, and growth characteristics on Congo red-magnesium oxalate agar appear to be helpful in identifying pathogenic strains [57].

Vibrio species. Twelve species are included in the genus Vibrio, and many are associated with gastroenteritis. In addition to the V cholerae 01 serogroups, which are known to cause classic cholera, a new non-01 serogroup, 0139, has now be­come an important cause of epidemic cholera. In addition to V cholerae, other species are associated with gastroenteritis; these species include V parahaemolyticus (the most common non-cholerae species) as well as V mimicus, V jiuvialis, and V hollisae (which can occasionally be detected in the labora­tory with use of selective media) [58].

Parasitic Agents

Many advances have been made in the diagnosis of parasitic diseases. Because of these advances, clinicians may wish to rethink the strategy of performing ova and parasite examina­tions on all samples and specifically request tests for particular agents.

Giardia lamblia, Cryptosporidium, and E. histolytica. G. lamblia is usually detected during a routine ova and parasite examination by examination of concentrated stool samples and permanent smears. In the past few years, several alternative methods for detecting G. lamblia in stool samples have been developed and commercialized, and these methods appear to have greatly improved sensitivity when compared with conven­tional methods.

Direct immunofluorescence staining (DF A), which is avail­able commercially, has been evaluated by a number ofinvesti­gators; it has been shown to significantly increase the sensitivity for detecting G. lamblia. Garcia et aL [59] found that DF A was 100% sensitive and specific for detecting G. lamblia in stool samples, compared with a conventional trichrome stain, which was 93% sensitive. In a more recent study by Alles et aL [60], DFA was compared with chlorazol black E stain and was found to be more sensitive for detecting G. lamblia. The sensitivity of immunofluorescence was 99.2%, while that of routine histologic staining was 66.4%. Both methods were 100% specific.

For Cryptosporidium, the numbers were equally impressive. Garcia et aL [59] found that DFA was more sensitive than the modified acid-fast stain (100% sensitivity vs. 92% sensitivity, respectively). However, Alles et aL [60] detected 69.6% more positive samples with the reagent than with modified acid-

fast stain. Immunofluorescence was 93% sensitive, while the modified acid-fast stain was 54.8% sensitive. A number of commercial antigen EIA detection systems have also been de­scribed for detecting G. lamblia and Cryptosporidium antigen in stool samples [61-63], and these also appear to be more sensitive than conventional microscopic analysis.

E. histolytica was recently divided into two species: the pathogenic strain retains the name E. histolytica, and the non­pathogenic species has been renamed E. dispar [64, 65]. Al­though the two species were formerly identified as a single organism on the basis of microscopic findings, they can be distinguished by the presence of a specific lectin in the patho­genic strains. Haque et aL [65] recently evaluated a commercial EIA specific for detection of pathogenic E. histolytica in stool samples. When compared with culture, microscopic examina­tion was 60% sensitive and 79% specific, and EIA was 80% sensitive and 99% specific. When compared with zymodeme analysis to differentiate pathogenic from nonpathogenic strains, the EIA was 95% sensitive and 93% specific. Thus, EIA has the potential for improved diagnosis of E. histolytica infection; however, additional studies are needed to better assess this test's performance.

The results of these studies suggest that these newer methods are superior to conventional parasitological microscopy and that they should be used more routinely for diagnosis. For patients in whom giardiasis, amebiasis, or cryptosporidiosis is suspected, specific tests as described above should be requested rather than the conventional microscopic examination for ova and parasites. An expanded menu of parasitological tests that allow more-specific options for testing should now be available to clinicians. Since conventional microscopic parasitological examinations are among the most time-consuming procedures in the laboratory, the use of some ofthe specific tests described above may lead to more cost-effective and clinically effective use of the parasitology laboratory.

Microsporidia. Microsporidia are a group of protozoa that have emerged as a cause of diarrheal disease, predominantly in patients with AIDS. However, the role of these organisms in the pathogenesis of gastrointestinal disease is unclear [66]. Four genera are in this group and include Enterocytozoon, Encephalitozoon, Pleistophora, and Nosema; all but Nosema have been associated with diarrhea in patients with AIDS. A review of the associated pathology and detection of these or­ganisms was recently published [67]. Detection of these organ­isms in the clinical laboratory poses significant challenges be­cause of their small size and difficulty in distinguishing them from artifacts with use of the available staining techniques.

Several light or fluorescent microscopic stains are currently available in laboratories, most of which use Weber's modified trichrome stain [68, 69]; more recently, Uvitex 2B, a fluorescent stain, has shown some promise [69]. For the accurate identifi­cation of micro sporidia, strict adherence to the diagnostic crite­ria of size, shape, and morphological characteristics is critical

CID 1996;23 (December) Microbiological Diagnosis of Diarrhea 1299

[69]. Molecular biological techniques such as PCR may ulti­mately provide more practical and specific methods in the near future [70]. Electron microscopy is the definitive procedure for identification of microsporidia to the species level, and species identification may be important in the future if antiprotozoal therapy is found to be effective [71].

Blastocystis hominis. The role of B. hominis in causing diar­rheal disease is unclear [72]. When this organism is present in large numbers in the stool, in the absence of other identifiable etiologic agents, patients have been reported to have diarrhea, abdominal pain, nausea, and vomiting that have responded to therapy with agents such as metronidazole [73-75]. The organ­ism is readily identified with use of routine microscopic methods.

Fungal Agents

It has been suggested that Candida species may be a cause of antibiotic-associated diarrheal disease; however, only a limited number of patients have been studied [76, 77]. The role ofthe laboratory in detecting Candida in stool samples is unclear at this time. At the very least, testing should not be performed on a routine basis.

Viral Agents

It is estimated that 30%-40% of episodes of infectious diar­rhea in the United States are caused by viral agents [78]. Rota­virus is the most important and common cause of serious gas­troenteritis; this agent causes significant morbidity that results in the need for hospitalization, particularly among children [78]. In addition, rotavirus frequently causes nosocomial infec­tions, and timely diagnosis may help limit the spread of infec­tion. A large number of detection systems are commercially available and include EIA and rapid latex agglutination tests; all of these systems have different performance characteristics [79].

Less common causes of viral gastrointestinal infection in­clude adenoviruses (types 40 and 41), caliciviruses (including Norwalk virus), astroviruses, and coronaviruses [80]. Commer­cial EIAs are available for detecting enteric adenoviruses [81]. However, at this time electron microscopy remains the only practical method for detecting the other viruses. Immunologic and molecular approaches for detecting these other viral agents have been published but remain in the developmental stages.

References

1. Snyder JD, Merson MH. The magnitude of the global problem of acute

diarrhoeal disease: a review of active surveillance data. Bull WHO 1982;60:605-13.

2. Lew JF, Glass RI, Gangarosa RE, Cohen IP, Bern C, Moe CL. Diarrheal

deaths in the United States, 1979 through 1987: a special problem for the elderly. JAMA 1991;265:3280-4.

3. Valenstein P, Pfaller M, Yungbluth M. The use and abuse of routine stool

microbiology: a College of American Pathologists Q-probes study of 601 institutions. Arch Pathol Lab Med 1996; 120:206-11.

4. Koplan JP, Fineberg HV, Ferraro MJB, Rosenberg ML. Value of stool

cultures. Lancet 1980;2:413-6.

5. Guerrant RL, Shields DS, Thorson SM, Schorling JB, Griischel DHM. Evaluation and diagnosis of acute infectious diarrhea. Am J Med 1985; 78(suppl 6B):91-8.

6. Siegel DL, Edelstein PH, Nachamkin I. Inappropriate testing for diarrheal

diseases in the hospital. JAMA 1990;263:979-82.

7. Ericsson CD, DuPont HL. Traveler's diarrhea: approaches to prevention and treatment. Clin Infect Dis 1993; 16:616-26.

8. Goodman LJ, Trenholme GM, Kaplan RL, et al. Empiric antimicrobial

therapy of domestically acquired acute diarrhea in urban adults. Arch Intern Med 1990; 150:541-6.

9. Mattila L, Peltola H, Siitonen A, Kyriinseppa H, Simula I, Kataja M. Short-term treatment of traveler's diarrhea with norfioxacin: a double­

blind, placebo-controlled study during two seasons. Clin Infect Dis 1993; 17:779-82.

10. Wistriim J, Jertborn M, Ekwall E, et al. Empiric treatment of acute diarrheal disease with norfioxacin: a randomized, placebo-controlled study. Ann Intern Med 1992; 117:202-8.

II. Wistriim J, Norrby SR. Fluoroquinolones and bacterial enteritis, when and for whom? J Antimicrob Chemother 1995;36:23-39.

12. Pichler RET, Diridl G, Stickler K, WolfD. Clinical efficacy of ciprofloxa­cin compared with placebo in bacterial diarrhea. Am J Med 1987; 82(suppI4A):329-32.

13. Rosenthal SL. Exacerbation of Sahnonella enteritis due to ampicillin. N Engl J Med 1969;280:147-8.

14. Neill MA, Opal SM, Heelan J, et al. Failure of ciprofloxacin to eradicate convalescent fecal excretion after acute salmonellosis: experience dur­ing an outbreak in health care workers. Ann Intern Med 1991; 114: 195-9.

15. Kuschner RA, Trofa AF, Thomas RJ, et al. Use of azithromycin for the treatment of Campylobacter enteritis in travelers to Thailand, an area where ciprofloxacin resistance is prevalent. Clin Infect Dis 1995;21: 536-4l.

16. Morris AJ, Wilson ML, Reller LB. Application of rejection criteria for stool ovum and parasite examinations. J Clin Microbiol 1992;30: 3213-6.

17. Fan K, Morris AJ, Reller LB. Application of rejection criteria for stool cultures for bacterial enteric pathogens. J Clin MicrobioI1993;31:2233-5.

18. Church DL, Cadrain G, Kabani A, Jadavji T, Trevenen C. Practice guide­

lines for ordering stool cultures in a pediatric population. Am J Clin Pathol 1995; 103:149-53.

19. Yannelli B, Gurevich I, Schoch PE, Cunha BA. Yield of stool cultures, ova and parasite tests, and Clostridium difficile determinations in nosocomial diarrheas. Am J Infect Control 1988; 16:246-9.

20. Morris AJ, Murray PR, Reller LB. Contemporary testing for enteric patho­gens: the potential for cost, time, and health care savings. J Clin Micro­bioi 1996;34:1776-8.

21. Boyce TG, Pemberton AG, Wells JG, Griffin PM. Screening for Esche­richia coli 0157 :H7 -a nationwide survey of clinical laboratories. J Clin MicrobioI1995;33:3275-7.

22. Nachamkin I. Laboratory diagnosis of bacterial gastroenteritis. Adv Pathol Lab Med 1994; 7:259-71.

23. Brownlee HJ Jr, ed. Management of acute nonspecific diarrhea. Am J Med 1990; 88(suppl 6A): 1 S-37S.

24. Johanson JF, Sonnenberg A. Efficient management of diarrhea in the acquired immunodeficiency syndrome (AIDS). Ann Intern Med 1990; 112:942-8.

25. Manabe YC, Vinetz JM, Moore RD, Merz C, Charache P, Bartlett JG. Clostridium difficile colitis: an efficient clinical approach to diagnosis. Ann Intern Med 1995; 123:835-40.

1300 Hines and Nachamkin em 1996;23 (December)

26. Cooper GS, Lederman MM, Salata RA. A predictive model to identify

Clostridium difficile toxin in hospitalized patients with diarrhea. Am J

GastroenteroI1996;91:80-4.

27. Siegel D, Cohen PT, Neighbor M, et al. Predictive value of stool examina­

tion in acute diarrhea. Arch Pathol Lab Med 1987; 111:715-8.

28. Nolte FS. Practical considerations in the laboratory diagnosis of bacterial enteric infections. Am J Clin Patho11994; IOI(suppl):SI4-7.

29. Guerrant RL. Principles and syndromes of enteric infection. In: Mandell

GL, Bennett JE, Dolin R, eds. Mandell, Douglas and Bennett's princi­ples and practice of infectious diseases. 4th ed. Vol. 1. New York:

Churchill Livingstone, 1995:945-62.

30. Choi SW, Park CH, Silva TMJ, Zaenker EI, Guerrant RL. To culture or

not to culture: fecal lactoferrin screening for inflammatory bacterial diarrhea. J Clin MicrobioI1996;34:928-32.

31. Donowitz M, Kokke FT, Saidi R. Evaluation of patients with chronic

diarrhea. N Engl J Med 1995;332:725-9. 32. Tarr PI, Clausen CR, Christie DL. Bacterial and protozoal gastroenteritis

[letter]. N Engl J Med 1992;326:489.

33. Tarr PI, Neill MA, Clausen CR, Watkins SL, Christie DL, Hickman RO.

Escherichia coli 0157:H7 and the hemolytic uremic syndrome: impor­tance of early cultures in establishing the etiology. J Infect Dis 1990;

162:553-6.

34. Harris JC, DuPont HL, Hornick RB. Fecal leukocytes in diarrheal illness.

Ann Intern Med 1972;76:697-703.

35. Taylor DN, Echeverria P, Pal T, et al. The role of Shigella spp., enteroinva­sive Escherichia coli, and other enteropathogens as causes of childhood

dysentery in Thailand. J Infect Dis 1986; 153:1132-8.

36. Y ong WH, Mattia AR, Ferraro MJ: Comparison of fecallactoferrin latex

agglutination assay and methylene blue microscopy for detection of

fecal leukocytes in Clostridium difficile-associated disease. J Clin Mi­

crobiol 1994; 32: 1360-1.

37. Marx CE, Morris A, Wilson ML, Reller LB. Fecal leukocytes in stool

specimens submitted for Clostridium difficile toxin assay. Diagn Micro­

bioI Infect Dis 1993; 16:313-5.

38. Novak R, Sadowski L, Klespies SL, Igel Hl How useful are fecal neutro­phil determinations? Laboratory Medicine 1995;26:743-5.

39. Silletti RP, Lee G, Ailey E. Role of stool screening tests in diagnosis of

inflammatory bacterial enteritis and in selection of specimens likely to

yield invasive enteric pathogens. J Clin MicrobioI1996;34:1161-5.

40. Nachamkin 1. Campylobacter and Arcobacter. In: Murray PR, Baron EJ,

Pfaller MA, Tenover FC, Yolken RH, eds. Manual of clinical microbiol­

ogy. 6th ed. Washington, DC: ASM Press, 1995:483-91.

41. Marti H, Koella JC. Multiple stool examinations for ova and parasites and

rate of false-negative results. J Clin Microbiol 1993; 31 :3044-5.

42. Renshaw AA, Stelling 1M, Doolittle MH. The lack of value of repeated

Clostridium difficile cytotoxicity assays. Arch Pathol Lab Med 1996; 120:49-52.

43. Persing DH, Smith TF, Tenover FC, White TJ, eds. Diagnostic molecular

microbiology: principles and applications. Washington, DC: American

Society for Microbiology, 1993.

44. Janda 1M, Abbott SL, Carnahan AM. Aeromonas and Plesiomonas. In:

Murray PR, Baron EJ, Pfaller MA, Tenover FC, Yolken RH, eds. Man­

ual of clinical microbiology. 6th ed. Washington, DC: ASM Press, 1995:

477-82.

45. Mishu B, Blaser Ml Role of infection due to Campylobacter jejuni in the initiation of Guillain-Barre syndrome. Clin Infect Dis 1993; 17: 104-8.

46. Peterson LR, Kelly Pl The role of the clinical microbiology laboratory

in the management of Clostridium difficile-associated diarrhea. Infect

Dis Clin North Am 1993;7:277-93.

47. Gumerlock PH, Tang YJ, Silva J Jr. PCR detection of toxigenic Clostrid­

ium difficile. In: Persing DH, Smith TF, Tenover FC, White TJ, eds.

Diagnostic molecular microbiology: principles and applications. Wash­

ington, DC: American Society for Microbiology, 1993:288-93.

48. Nachamkin I, Lotz-Nolan L, Skalina D. Evaluation of a commercial cyto­

toxicity assay for detection of Clostridium difficile toxin. J Clin Micro­

bioi 1986;23:954-5.

49. Gilligan PH, Walden TP, Kelly WF, Wait KJ, Kraft JA, Willis DH. The use

of a commercially available enzyme innnunoassay for the detection of Clostridium difficile toxin A. Arch Pathol Lab Med 1993; 117:507-10.

50. Doern GV, Coughlin RT, Wu L. Laboratory diagnosis of Clostridium

difficile-associated gastrointestinal disease: comparison of a mono­clonal antibody enzyme innnunoassay for toxins A and B with a mono­

clonal antibody enzyme innnunoassay for toxin A only and two cytotox­

icity assays. J Clin MicrobioI1992;30:2042-6.

51. Shanholtzer CJ, Willard KE, Holter JJ, Olson MM, Gerding DN, Peterson

LR. Comparison ofthe VIDAS Clostridium difficile toxin A immunoas­say with C. difficile culture and cytotoxin and latex tests. J Clin Micro­

bioI 1992;30:1837-40.

52. Gray LD. Escherichia, Salmonella, Shigella, and Yersinia. In: Murray PR,

Baron EJ, Pfaller MA, Tenover FC, Yolken RH, eds. Manual of clinical microbiology. 6th ed. Washington, DC: ASM Press, 1995:450-6.

53. Calderwood SB, Acheson DWK, Keusch GT, et al. Proposed new nomen­clature for SLT (VT) family. ASM News 1996;62:118-9.

54. Ritchie M, Partington S, Jessop J, Kelly MT. Comparison of a direct

fecal shiga-like toxin assay and sorbitol-MacConkey agar culture for

laboratory diagnosis of enterohemorrhagic Escherichia coli infection. J

Clin MicrobioI1992;30:461-4.

55. Ramotar K, Waldhart B, Church D, Szumski R, Louie Tl Direct detection

of verotoxin-producing Escherichia coli in stool samples by PCR. J Clin MicrobioI1995;33:519-24.

56. Dylla BL, Vetter EA, Hughes JG, Cockerill FR. Evaluation of an immuno­

assay for direct detection of Escherichia coli 0157 in stool specimens.

J Clin MicrobioI1995;33:222-4.

57. Farmer JJ III, Carter GP, Miller VL, Falkow S, Wachsmuth IK. Pyrazinam­

idase, CR-MOX agar, salicin fermentation-esculin hydrolysis, and d­

xylose fermentation for identifying pathogenic serotypes of Yersinia

enterocolitica. J Clin MicrobioI1992;30:2589-94.

58. McLaughlin Jc. Vibrio. In: Murray PR, Baron EJ, Pfaller MA, Tenover FC, YolkenRH, eds. Manual of clinical microbiology. 6th ed. Washing­

ton, DC: ASM Press, 1995:465-76.

59. Garcia LS, Shum AC, Bruckner DA. Evaluation of a new monoclonal

antibody combination reagent for direct fluorescence detection of Giar­

dia cysts and Cryptosporidium oocysts in human fecal specimens. J

Clin MicrobioI1992;30:3255-7.

60. Alles AJ, Waldron MA, Sierra LS, Mattia AR. Prospective comparison

of direct immunofluorescence and conventional staining methods for

detection of Giardia and Cryptosporidium spp. in human fecal speci­

mens. J Clin MicrobioI1995;33:1632-4.

61. Rosoff JD, Sanders CA, Sonnad SS, et al. Stool diagnosis of giardiasis

using a commercially available enzyme immunoassay to detect Gi­

ardia-specific antigen 65 (GSA 65). J Clin Microbiol 1989; 27:

1997-2002.

62. Parisi MT, Tierno PM Jr. Evaluation of a new rapid commercial enzyme

innnunoassay for detection of Cryptosporidium oocysts in untreated

stool specimens. J Clin MicrobioI1995;33:1963-5.

63. Zinnnerman SK, Needham CA. Comparison of conventional stool concen­

tration and preserved-smear methods with Merifluor Cryptosporidium/

Giardia direct innnunofluorescence assay and ProSpecT Giardia EZ microplate assay for detection of Giardia lamblia. J Clin Microbiol

1995;33:1942-3.

64. Allason-Jones E, Mindel A, Sargeaunt P, Williams P. Entamoeba histolyt­

ica as a commensal intestinal parasite in homosexual men. N Engl J

Med 1986;315:353-6.

65. Haque R, Neville LM, Hahn P, Petri WA Jr. Rapid diagnosis of Entamoeba

infection by using Entamoeba and Entamoeba histolytica stool antigen

detection kits. J Clin MicrobioI1995;33:2558-61.

cm 1996;23 (December) Microbiological Diagnosis of Diarrhea 1301

66. Rabeneck L, Gyorkey F, Genta RM, Gyorkey P, Foote LW, Risser JMH. The role of Microsporidia in the pathogenesis of HIV-related chronic diarrhea. Ann Intern Med 1993; 119:895-9.

67. Schwartz DA, Sobottka I, Leitch GJ, Cali A, Visvesvara GS. Pathology of microsporidiosis: emerging parasitic infections in patients with acquired immunodeficiency syndrome. Arch Pathol Lab Med 1996; 120:173-88.

68. Weber R, Bryan RT, Owen RL, et al. Improved light-microscopical detec­tion of microsporida spores in stool and duodenal aspirates. N Engl J Med 1992;326:161-6.

69. DeGirolami PC, Ezratty CR, Desai G, et al. Diagnosis of intestinal mi­crosporidiosis by examination of stool and duodenal aspirate with We­ber's modified trichrome and Uvitex 2B stains. J Clin Microbiol1995; 33:805-10.

70. Fedorko DP, Nelson NA, Cartwright CPo Identification of microsporida in stool specimens by using PCR and restriction endonucleases. J Clin MicrobioI1995;33:1739-41.

71. Sobottka I, Albrecht H, Schafer H, et al. Disseminated Encephalitozoon (Septataj intestinalis infection in a patient with AIDS: novel diagnostic approaches and autopsy-confirmed parasitological cure following treat­ment with albendazole. J Clin MicrobioI1995;33:2948-52.

72. Garcia LS, Bruckner DA. Intestinal protozoa: amebae. In: Garcia LS, Bruckner DA, eds. Diagnostic medical parasitology. 2nd ed. Washing­ton, DC: American Society for Microbiology; 1993:6-30.

73. Zierdt CH. Blastocystis hominis-past and future. Clin Microbiol Rev 1991;4:61-79.

74. Doyle PW, Helgason MM, Mathias RG, Proctor EM. Epidemiology and pathogenicity of Blastocystis hominis. J Clin Microbiol 1990;28: 116-21.

75. Qadri SMH, AI-Okaili GA, AI-Dayel F. Clinical significance of Blasto­cystis hominis. J Clin MicrobioI1989;27:2407-9.

76. Danna PL, Urban C, Bellin E, Rahal JJ. Role of candida in pathogenesis of antibiotic-associated diarrhoea in elderly inpatients. Lancet 1991; 337:511-4.

77. Gupta TP, Ehrinpreis MN. Candida-associated diarrhea in hospitalized patients. Gastroenterology 1990;98:780-5.

78. Blacklow NR, Greenberg HB. Viral gastroenteritis. N Engl J Med 1991; 325:252-64.

79. Christensen ML. Rotaviruses. In: Murray PR, Baron EJ, Pfaller MA, Ten­over FC, Yolken RH, eds. Manual of clinical microbiology. 6th ed. Washington, DC: ASM Press, 1995:1012-6.

80. Petric M. Caliciviruses, astroviruses, and other diarrheic viruses. In: Mur­ray PR, Baron EJ, Pfaller MA, Tenover FC, Y olken RH, eds. Manual of clinical microbiology. 6th ed. Washington, DC: ASM Press, 1995: 1017-24.

81. Hierholzer JC. Adenoviruses. In: Murray PR, Baron EJ, Pfaller MA, Ten­over FC, Yolken RH, eds. Manual of clinical microbiology. 6th ed. Washington, DC: ASM Press, 1996:947-55.