Point-Counterpoint: What Is the Optimal Approach for Detection of Clostridium difficile Infection?

INTRODUCTION

In 2010, we published an initial Point-Counterpoint on the laboratory diagnosis of Clostridium difficile infection (CDI). At that time, nucleic acid amplification tests (NAATs) were just becoming commercially available, and the idea of algorithmic approaches to CDI was being explored. Now, there are numerous NAATs in the marketplace, and based on recent proficiency test surveys, they have become the predominant method used for CDI diagnosis in the United States. At the same time, there is a body of literature that suggests that NAATs lack clinical specificity and thus inflate CDI rates. Hospital administrators are taking note of institutional CDI rates because they are publicly reported. They have become an important metric impacting hospital safety ratings and value-based purchasing; hospitals may have millions of dollars of reimbursement at risk. In this Point-Counterpoint using a frequently asked question approach, Ferric Fang of the University of Washington, who has been a consistent advocate for a NAAT-only approach for CDI diagnosis, will discuss the value of a NAAT-only approach, while Christopher Polage of the University of California Davis and Mark Wilcox of Leeds University, Leeds, United Kingdom, each of whom has recently written important articles on the value of toxin detection in the diagnosis, will discuss the impact of toxin detection in CDI diagnosis.

FREQUENTLY ASKED QUESTIONS

1. Why is there so much controversy about the performance of Clostridium difficile diagnostic tests?

Fang. Diagnostic tests detect either toxigenic C. difficile or its toxins. Many labs have switched from toxin assays to nucleic acid amplification tests (NAATs) that detect toxigenic C. difficile in order to maximize sensitivity, as toxin assays were previously missing cases of clinically significant C. difficile infection (CDI). However, some recent studies have highlighted that NAATs can be positive for colonized patients without disease, and patients with positive toxin assays may have a worse prognosis than those with a positive NAAT only (1, 2). This has renewed controversy about the optimal approach to diagnosis of CDI.

Polage and Wilcox. The performance of C. difficile diagnostic tests is controversial for 4 reasons.

(i) There is no reliable clinical or laboratory definition for CDI that accurately distinguishes true CDI from non-CDI-related symptoms in all patients (3). Most diarrhea in hospitals is not due to CDI, and virtually all clinical signs and symptoms of CDI are nonspecific and occur commonly in patients without CDI (4, 5). Asymptomatic C. difficile colonization is also common in hospitals, particularly among patients who get selected for C. difficile testing due to risk factors shared between colonized patients and those with CDI (6, 7). Thus, the positive predictive value of detecting toxigenic C. difficile in routine diarrheal samples submitted to the laboratory is low and insufficient to diagnose CDI (1–3, 7).

(ii) The measured performance of C. difficile diagnostic tests is highly dependent on the definition of CDI and the ratio of CDI to colonization in the population being tested (2, 3, 8). For example, toxin tests are sensitive (and agreement with toxigenic culture is high) in patients with pseudomembranous colitis due to the high ratio of CDI to colonization in this population (8). Conversely, toxin tests appear less sensitive in routine stool samples submitted to the laboratory due to frequent overlap of non-CDI diarrhea with C. difficile colonization and the lower ratio of CDI to colonization in this population (1–3, 8, 9).

(iii) Anecdotal experiences with cases of severe CDI missed by toxin tests have promoted a desire for absolute sensitivity, regardless of specificity, and an erroneous belief that all patients with toxigenic C. difficile and diarrhea have CDI as the cause of their symptoms (9–14). Widespread misclassification of non-CDI diarrhea in patients with C. difficile colonization as “CDI” has reinforced the belief that toxin tests are insensitive for CDI without systematic investigation to verify the true frequency of disease (2, 9, 11, 15–17).

(iv) C. difficile tests, including those with the same target, vary in performance accuracy; for example, there are marked and sometimes significant differences in sensitivity and specificity between commercial toxin detection tests (1, 3, 9). Thus, use of less well performing tests can reinforce perceptions driven by other factors (see above).

Editor's comment. The measured accuracy of any diagnostic test is dependent upon the reference test to which the diagnostic test is being compared. The American Society for Microbiology has a group that is currently working on an evidence-based practice guideline for laboratory detection of C. difficile infection. There are over 15 different reference methods that have appeared in this literature, some of which are clearly biased. This lack of a standard reference method to define C. difficile infection clearly complicates an already very complicated literature, and there is no consensus in sight.

2. What are the effects of using nucleic acid amplification testing for C. difficile on C. difficile infection data that institutions report to public health authorities?

Fang. Since NAATs are more sensitive than toxin assays, the introduction of a NAAT will initially increase the apparent infection rate at an institution. However, this is mitigated by two factors. First, the National Health Safety Network applies a correction factor for institutions that use NAATs to diagnosis CDI, so that institutions using more-sensitive diagnostic methods will not be penalized (18). Second, the greater detection of toxigenic C. difficile by NAATs can facilitate more-effective infection control measures so that institutional infection rates subsequently decline (19–21). This has been the experience at my own institution, where several years ago our CDI rates fell within a few months of introducing NAATs and have remained low ever since. The sensitive detection of toxigenic C. difficile can facilitate efforts to reduce institutional transmission. That said, public health agencies must recognize that laboratory data alone cannot be used to accurately monitor CDI rates, as laboratory tests detect both colonized and infected patients.

Polage and Wilcox. When positive laboratory test results are used as the sole measure of health care facility-onset CDI, as is currently the case for most hospitals in the United States, NAAT-based CDI diagnosis can have a dramatic effect on the number of CDI cases that institutions report publically and affect hospital reimbursement under value-based payment programs (18, 22–24). This is because NAAT-based CDI testing results in public reporting of all fecal-toxin-negative samples with toxigenic C. difficile as positive regardless of clinical disease or treatment. Most hospitals using NAATs or a glutamate dehydrogenase (GDH) immunoassay plus a NAAT for CDI diagnosis see an increase in the number of “CDI cases” reported publically by 1.5- to 3-fold over rates derived from toxin tests (18, 23, 24). The NAAT-related increase is partially accounted for by an adjustment in the National Healthcare Safety Network (NHSN) standardized infection ratio (SIR) calculation used to compare hospital performance, but the current adjustment does not fully correct for the increased number of positive results at all hospitals (24). This might be appropriate if all toxin-negative patients with C. difficile detected by NAATs had CDI clinically, but this is not the case (2, 3, 8). Recent outcome studies show that most toxin-negative patients with C. difficile detected by NAATs or culture recover spontaneously without treatment and have a significantly lower rate of adverse events than toxin-positive patients; furthermore, the duration of symptoms for toxin-negative patients with C. difficile detected by NAATs is similar to that for C. difficile-negative control patients (1, 2, 25). These findings suggest that using a NAAT as a standalone test for CDI diagnosis results in a considerable amount of overdiagnosis that has important clinical, financial, and reputational implications for hospitals (2, 25). For this reason, guidelines in the United Kingdom and Europe now recommend toxin testing to confirm CDI in NAAT-positive patients and consideration of other causes for symptoms before diagnosis and treatment of CDI in toxin-negative patients (3).

3. Should GDH immunoassays be used as a screening test to determine which stool specimens should be subjected to toxin or nucleic acid amplification testing for C. difficile?

Fang. GDH immunoassays are more sensitive than toxin assays and can be used to screen specimens for the presence of C. difficile (26). However, GDH is expressed by both toxigenic and nontoxigenic strains of C. difficile, so GDH-positive specimens must be further tested using NAATs and/or toxin assays. Such an approach is less expensive than performing NAATs on all specimens but is also less sensitive, particularly for non-027 strains (27, 28). This is not because of strain-dependent differences in GDH expression but most likely because 027 strains tend to attain higher organism burdens. The calculated sensitivity of the GDH immunoassay is dependent on the sensitivity of the comparator method, and studies, including a blind multicenter trial using the most-sensitive comparators (NAAT and toxigenic culture with detection of both spores and vegetative cells), have shown that GDH assays miss approximately 20% of specimens detected by NAATs in patients with symptomatic CDI (17, 27, 28). In short, a GDH-based algorithm is less costly but sacrifices sensitivity.

Polage and Wilcox. GDH detection is sensitive for CDI because C. difficile vegetative cells express and secrete GDH extracellularly, and GDH may play a role in C. difficile colonization in vivo (29). As a result, most clinical samples with toxigenic C. difficile detectable by culture or NAATs are positive by GDH immunoassays, and virtually all samples with toxins detectable are positive for GDH (3, 9, 30). The occasional samples that are positive by NAATs but negative for GDH have a low concentration of C. difficile organisms and no toxins, suggesting that these are most likely C. difficile carriers or patients on treatment (30). Most laboratory comparisons find that GDH immunoassays are >90% sensitive for C. difficile, as confirmed by two meta-analyses; a few studies report slightly lower sensitivities in the range of 83.1 to 87.6% (3, 9, 26). In the most recent meta-analysis, the pooled sensitivities of GDH immunoassays were 94% (95% confidence interval [CI], 89 to 97%) and 96% (95% CI, 86 to 99%) relative to those of the cell cytotoxin neutralization assay and toxigenic culture, respectively; the pooled specificity was 90 to 96% (3). Finally, recent studies showed that GDH expression is a reliable characteristic of all common C. difficile strains, contradicting an earlier study, which hypothesized that differential GDH expression might explain the lower sensitivity of two-step immunoassay algorithms for some C. difficile ribotypes (9, 27). In summary, GDH immunoassays are less expensive and modestly less sensitive as a screening test than some NAATs; NAATs are generally more sensitive, specific, and expensive. Samples that test positive by either method should be retested by a fecal toxin A/B immunoassay to confirm clinical CDI disease (3). Individual laboratories should choose the C. difficile screening test and algorithm that works best in their lab and institution.

4. What is the most cost-effective strategy for C. difficile diagnosis?

Fang. Although immunoassay methods are less costly for the laboratory than NAATs, a recent cost-effectiveness analysis has determined that a NAAT is the most cost-effective approach from an institutional standpoint due to the $9,000 to $13,000 cost of each missed case of CDI (31). Another study found that patients diagnosed with CDI by NAATs had a 2-day-shorter median length of stay than patients diagnosed by a toxin immunoassay, even though the patients did not differ with regard to comorbidity, prior hospitalizations, laboratory parameters, or mortality (32). Length of stay is an important contributor to the financial costs of CDI (33, 34), and the authors suggested that a sensitive NAAT might result in more timely diagnosis and treatment (32). In addition, reliance on a less sensitive diagnostic method may lead to more empirical therapy (35) and repeat laboratory testing, because clinicians lack confidence in a negative result. Thus, the use of a NAAT can promote responsible antimicrobial stewardship and reduce unnecessary antibiotic and laboratory utilization.

Polage and Wilcox. The latest guidelines recommend a two- or three-step algorithm as the most effective strategy to diagnose CDI and minimize overdiagnosis of C. difficile-colonized individuals who have other causes for their diarrheal symptoms (3). The algorithm should start with a rapid and sensitive screening test with a high negative predictive value for CDI, such as a GDH immunoassay or NAAT, to minimize empirical isolation and treatment of non-CDI patients (3). Samples with a positive screening test should be retested with a toxin A/B immunoassay to identify patients with toxins, who have the highest likelihood of CDI clinically and need for treatment (3). Patients with toxigenic C. difficile but no fecal toxins need additional clinical evaluation to distinguish incidental C. difficile colonization (most patients) from CDI with a negative toxin test (fewer patients) (3). The overall sensitivity and specificity of this approach were verified in a multicenter prospective study in the United Kingdom and supported in a recent meta-analysis (1, 3). The emphasis on fecal toxin detection in this algorithm to identify patients with high (toxin-positive patients) and low (toxin-negative patients) likelihoods of clinical CDI disease is supported by outcome studies in multiple countries (1, 2, 8, 25). In terms of cost, new economic models are needed to determine which strategy is best since previous models inappropriately assumed that patients with toxigenic C. difficile and negative toxin tests had CDI and overlooked the costs of CDI overdiagnosis, including decreased hospital reimbursement (31, 36).

Editor's comment. A March 2016 survey of 70 members of Clinmicronet, a global listserv of doctoral clinical microbiologists, showed that 55 laboratories used a NAAT-only approach, while 9 used a GDH/toxin screen with PCR confirmation for GDH/toxin-discrepant specimens. College of American Pathologists (CAP) surveys of C. difficile testing also show a preponderance of laboratories using a NAAT-only approach. Only 6 of 70 respondents used the algorithm of a GDH or NAAT screen with toxin confirmation of the screen-positive results described by Polage and Wilcox. Three laboratories were considering changing to this approach. One microbiologist commented that the decision to change to this algorithm was driven by a hospital administration belief that using this approach would reduce reported CDI rates.

5. Why do studies of symptoms and clinical outcomes in patients who have C. difficile DNA or bacteria but not toxins in stool reach such different conclusions?

Fang. NAATs and culture-based methods are more sensitive but less specific than toxin assays, whereas toxin assays are less sensitive but more specific than NAATs. Thus, patient selection is critically important for the proper interpretation of test results. With regard to specificity, it is important to recognize that no C. difficile diagnostic assay is completely specific for clinical disease. Production of toxin is essential but not sufficient for disease, and even patients with high fecal toxin levels may be asymptomatic (37, 38), particularly if they have toxin-neutralizing antibodies (39). With regard to sensitivity, it is equally important to recognize that toxin assays can be negative in patients with symptomatic (and even life-threatening) CDIs (10, 13, 40, 41). The insensitivity of toxin assays has been demonstrated even for cases of pseudomembranous colitis and was a major consideration leading to the development of more-sensitive NAATs. In fact, a false-negative toxin assay is a risk factor for a fatal outcome in patients with fulminant CDI (10), and I note that one of the toxin-negative patients in the Polage et al. study (2) “had recurrent CDI as a contributing factor to death.” The bottom line is that a negative toxin assay cannot rule out the possibility of CDI. On the other hand, the greater sensitivity of NAAT or culture-based diagnostic methods can increase the likelihood of false-positive results, particularly in patients with a low clinical probability of C. difficile-associated disease. Exclusion of patients who fail to meet the clinical definition of diarrhea (or have formed stools), are receiving laxatives, or have previously tested positive can help to reduce the number of false-positive results. The best way to avoid false-positive test results is to restrict diagnostic testing to patients who have clinical presentations consistent with CDI, and inappropriate testing can account for many of the reported instances of “overdiagnosis” (1, 2). Institutional guidelines with clear criteria for diagnostic testing can be helpful in this regard.

Some have advocated the performance of both NAATs and toxin assays to optimize patient management. However, the data are conflicting. Although some studies suggest that patients with positive toxin assays have a worse prognosis than those with a positive NAAT only (1, 2), many other carefully conducted studies involving more than 2,000 patients have not found toxin assays to be predictive of symptoms, disease severity, mortality, transmissibility, or recurrence (11, 15, 16, 38, 42, 43). In any case, whether the detection of toxin is indicative of a worse prognosis is beside the point. The notion that a toxin assay can distinguish between colonization and infection is fundamentally flawed; the distinction between colonization and infection is a clinical one and cannot be based on laboratory assessment alone. As Dubberke and Burnham have noted, one must “treat the patient, not the test” (44). Some patients with positive toxin assays have asymptomatic colonization (37, 38), and some patients with negative toxin assays have CDIs (10, 11, 13, 15, 16, 40–43). More than half of patients with symptomatic CDIs would be missed by reliance on a toxin immunoassay (11, 15, 16, 42, 43), an unacceptably high proportion of false-negative results. Furthermore, patients with NAAT-positive/toxin-negative specimens may convert to toxin positivity on retesting; this was observed in 21% of individuals undergoing retesting in the Polage et al. study (2). I recommend using a negative NAAT to rule out the possibility of CDI and a positive NAAT to indicate the possibility of CDI in a patient with a compatible clinical presentation; using this approach, toxin assays are unnecessary. Treatment decisions should be based on clinical assessment and the presence or absence of toxigenic C. difficile, not on the ability or failure to detect fecal toxin.

I feel compelled to point out a self-contradiction in the European guidelines that advocate toxin testing. On one hand, the guidelines acknowledge that “the decision to treat CDI is ultimately a clinical decision” and that “treatment should not be withheld on the basis of laboratory tests alone,” but on the other hand, they state that “using NAAT as a stand-alone test and relying on clinical symptoms to discern patients with CDI from asymptomatic carriers is not an optimal approach” and that “samples with a positive first test result should be tested further with a toxin A/B EIA” (enzyme immunoassay) (3). On what should treatment decisions be based, clinical assessment or the presence of toxin? No wonder clinicians are confused.

I strongly disagree with the suggestion that a negative toxin assay means that a patient is only colonized and not infected (1); such a simplistic approach is likely to result in the underdiagnosis of CDIs and harm to patients. Although some suggest that symptomatic patients with CDIs and negative toxin assays have self-limited disease that will resolve without treatment (1, 2), this cannot be concluded from the available studies, as many of the patients in these studies who had negative toxin assays received empirical treatment for CDI. Furthermore, important clinical endpoints other than mortality, such as the duration and severity of symptoms, were not measured, and the length-of-stay for culture-positive and toxin-negative patients was actually significantly longer than that of controls for whom both tests were negative (1). Quite simply, the safety of withholding antimicrobial treatment from symptomatic patients with positive NAATs and negative toxin assay results has not been established. Untreated patients will also continue to shed C. difficile with the potential to transmit infection to others, in contrast to those receiving specific antimicrobial treatment (45).

Polage and Wilcox. There is a growing consensus that most patients with C. difficile DNA or bacteria but no fecal toxins (i.e., those who are toxin negative and C. difficile positive) are clinically distinct from toxin-positive patients, have better outcomes, and generally do not have CDI as a cause of their symptoms (1–3, 25). Overall, 14 of 18 studies (78%) have reported a clinical symptom or outcome difference in toxin-negative, C. difficile-positive patients, and large studies from multiple countries have found less-severe disease, a shorter duration of diarrhea, fewer CDI-related complications, and/or lower mortality in these patients (1, 2, 8, 11, 15–17, 25, 42, 43, 46–53). In several studies, outcomes were similar to those of negative controls despite delayed reporting or nonreporting of NAAT or culture results and delayed or no treatment for CDI, further supporting an alternate cause of symptoms (not CDI) (1, 2, 8, 46, 52).

Nonetheless, some studies reach the opposite conclusion, that toxin-negative, C. difficile-positive patients have CDI and are not different from toxin-positive patients, and it is important to understand how and why this might occur (11, 15–17, 42, 48). Most of these studies were not adequately designed or powered to detect a statistical difference in rare clinical outcomes, such as CDI-related complications or mortality, and erroneously interpret a nonsignificant P value as evidence that differences do not exist (a type II statistical error) (11, 15–17, 48). Many of these studies also have significant sources of bias, which likely contributed to the authors' conclusions, including clinical reporting or reviewer knowledge of NAAT results and automatic classification of patients with a positive NAAT or culture as having CDI regardless of disease status (11, 15–17, 42, 48). Another common problem is failure to acknowledge that many clinical signs and outcomes seen in patients tested for CDI are common and nonspecific in hospitals and so are not necessarily indicative of, or related to, CDI (e.g., diarrhea, leukocytosis, intensive-care unit care) (11, 16, 48). Preanalytic issues can also cause negative results. One study routinely placed fecal samples in Cary-Blair transport medium before toxin testing, making it likely that preanalytic dilution contributed to negative toxin EIA results and so masked the relationship between fecal toxins and CDI-related outcomes (42). In summary, there are good explanations for why some studies fail to find differences between toxin-positive and toxin-negative, C. difficile-positive patients, and understanding how and why such misinterpretations occur is critical to interpreting the literature in this controversial field.

Editor's comment. Because of the uncertainty of which testing approach is most accurate in predicting that a patient has CDI, it is clear that preanalytic considerations are essential in determining who should be tested for CDI. Ensuring that tested patients have documented diarrheal disease and have not received laxatives in the past 48 h is essential for diagnostic accuracy, regardless of testing approach.

6. Will increasing the sensitivity of assays for C. difficile toxins in stool increase the accuracy of toxin assays?

Fang. Not necessarily. Toxin assays with increased sensitivity may reduce the incidence of false-negative results. However, C. difficile toxins are labile at body temperature and susceptible to inactivation by digestive enzymes (54, 55), so a completely sensitive toxin-based assay may not be feasible. Even recent “ultrasensitive” toxin assays are still less sensitive than NAATs (56). The likelihood of clinical illness in individuals with positive NAAT and negative ultrasensitive toxin assay results remains to be determined. It should also be noted that improvements in the sensitivity of toxin assays will not solve the issue of false-positive results (i.e., specificity), which can be seen with any C. difficile diagnostic method.

Polage and Wilcox. Maybe. Higher sensitivity toxin assays will decrease the number of CDI cases “missed” by toxin tests and bring the analytical and clinical performance closer to that of the traditional cell cytotoxin neutralization assay (2, 30, 56, 57). This should be a good thing. However, lowering the threshold for positive results will also decrease the specificity for CDI and lead to classifying patients with transient or low levels of toxin due to C. difficile colonization and antibiotic exposure as (likely erroneously) having disease (2, 56, 57). It is not known whether detecting and treating these additional patients labeled as having CDI is necessary or beneficial (or possibly harmful), since most resolve their symptoms with minimal or no treatment (2). These issues could be addressed by quantifying the level of toxins to help physicians determine the likelihood that each patient has disease and warrants treatment (56, 57). In any case, the overall diagnostic accuracy will depend on the test performance characteristics in the population being tested. Test performance and diagnostic accuracy are affected by many factors, including local testing practices, use of diarrheagenic medications, and the prevalence of CDI, C. difficile carriage, non-CDI diarrhea, anti-toxin antibodies, and individual C. difficile strains in the population (5, 7, 58). Thus, high-sensitivity toxin tests will probably improve diagnostic accuracy in hospitals/units with good C. difficile testing practices, a low prevalence of C. difficile carriage, and a low prevalence of non-CDI diarrhea. However, diagnostic accuracy could easily be worse in hospitals/units with indiscriminant C. difficile testing and a high prevalence of C. difficile carriage and non-CDI diarrhea. Overall, accurate diagnosis of CDI depends on a multitude of factors and starts at the bedside, with good clinical evaluation of the likelihood of CDI and non-CDI diarrhea and appropriate sampling and testing. Having a high-sensitivity toxin test will definitely be an improvement but will not remove the need for laboratories to work with clinicians and nurses to optimize clinical evaluation, testing, and diagnosis of symptomatic patients.

7. Should the diagnostic testing strategy for C. difficile infection be different in oncology, transplant, and other immunocompromised patients?

Fang. Immunocompromised hosts are at increased risk for CDI, and at least some studies suggest comparable clinical severities of CDI in immunocompromised patients with positive toxin assays and those with a positive NAAT only (15, 48). However, as I advocate the use of a NAAT to diagnosis CDI in all patients, immunocompromised patients do not require a special testing approach.

Polage and Wilcox. No. The two-step algorithm recommended in European guidelines is still preferred in oncology, transplant, and immunocompromised patients (3). Moreover, diagnostic strategies based solely on detection of toxigenic C. difficile (e.g., a NAAT only) are likely to perform worse in these patients due to high rates of treatment-related diarrhea and C. difficile carriage (5, 59). The lower positive predictive value of detecting toxigenic C. difficile when diarrheal symptoms occur in these patients reinforces the need for judicious testing, thoughtful clinical evaluation, and fecal toxin testing to maximize the accuracy of CDI diagnoses in these groups (3, 5, 59).

8. What is the significance of asymptomatic carriage of toxigenic C. difficile?

Fang. Asymptomatic colonized patients are an important source of C. difficile transmission (6, 60) and are at substantially increased personal risk for the eventual development of symptomatic CDIs (61, 62). Therefore, the identification of asymptomatic carriers can enhance infection control and prevention efforts. A recent study suggests that detection and isolation of colonized patients can prevent hospital-acquired CDI (63), and a CDC analysis has concluded that reduced transmission due to the isolation of carriers was responsible for the reduction in CDI incidence (64). High-risk antibiotics (e.g., cephalosporins, fluoroquinolones, clindamycin) should be avoided if at all possible in patients known to carry toxigenic C. difficile, and the possibility of CDI should be immediately considered if diarrhea, fever, or other compatible symptoms develop.

Polage and Wilcox. Asymptomatic C. difficile carriers outnumber CDI patients by at least 5 to 1 in most hospitals and are likely to be an important source of nosocomial C. difficile transmission and infection (6, 7, 61, 63). A few studies have linked asymptomatic carriers to a third or more of hospital-onset CDI cases (6, 7, 60). These observations have sparked an interest in screening and isolating asymptomatic carriers as a strategy to decrease health care-associated CDI (6, 7, 63). So far, a single before-and-after study has been published, with results suggesting that screening may be effective (63). However, the current absence of proven interventions for asymptomatic colonization and potential ramifications of isolating large numbers of patients emphasize the need for larger, well-controlled, multicenter studies to confirm the effectiveness of screening before widespread adoption (7, 63).

Asymptomatic C. difficile colonization may also be an important predisposing risk factor for CDI, but the story is somewhat mixed (58, 61, 65). Studies from the 1990s associated lack of symptoms after C. difficile acquisition with preexisting antitoxin antibodies and prior asymptomatic C. difficile colonization with a lower risk of CDI in hospitals (58, 65). These studies promoted the belief that most asymptomatic C. difficile carriers were immune to C. difficile toxins but the high rate of colonization with a nontoxigenic C. difficile strain (which also protects against CDI) was a potential confounder in one often-mentioned review (58, 65). More recently, asymptomatic C. difficile colonization has been associated with an increased risk of CDI, but it is unclear if this is an artifact of NAAT-based testing, a change in the epidemiology and pathophysiology of CDI, or simply a reflection of differential risk according to the toxigenic status of colonizing strains (61). Hence, more work is needed to determine the relationship between asymptomatic C. difficile carriage and the subsequent risk of CDI.

Finally, as noted above, asymptomatic C. difficile colonization is probably an important source of erroneous CDI diagnoses in hospitals using C. difficile tests with poor predictive value for CDI, as colonized patients with diarrheal symptoms due to medications, underlying disease, and other infectious agents will yield positive (misleading) results (2, 5, 7, 66–68).

Editor's comment. One of the ongoing discussions concerning C. difficile is whether admission screening has any benefit. If asymptomatic patients are found to be colonized, they are likely to be isolated since there are data suggesting that colonized patients may spread C. difficile. Although limiting the use of “high-risk” antimicrobials in colonized patients is an attractive idea, whether it will reduce CDI rates is not understood. Since treatment does not reliably clear C. difficile in a significant proportion of patients with CDI, antimicrobial clearance of carriage is also likely to be ineffective as well.

9. Much of the debate seems to be about the potential for false-positive results for C. difficile infection. What are the consequences of administering antibiotics to treat C. difficile infection to patients who are colonized, but not infected, with C. difficile?

Fang. Administering antibiotics to asymptomatic colonized patients will not provide a clinical benefit and will disrupt the host microbiota. The use of unnecessary antibiotics can also promote the emergence of antibiotic-resistant organisms, such as vancomycin-resistant enterococci (VRE) (69).

Polage and Wilcox. Antibiotic treatment for CDI is not benign. Metronidazole and vancomycin increase the risk of colonization and infection with multidrug-resistant organisms and promote rebound overgrowth of C. difficile in colonized patients after antibiotic discontinuation, which can lead to prolonged shedding or active infection (CDI) (70–72). Reflexive treatment of patients with false-positive results for CDI can also lead to delayed recognition of outbreaks (e.g., norovirus) or alternative diagnoses (e.g., medication-induced diarrhea, ischemic colitis) and treatment failure (66–68). In the near future, antibiotic use in hospitals will be reported publically, and hospitals will be mandated to implement antimicrobial stewardship programs to improve antibiotic use, creating additional incentives for hospitals to curb excessive/unnecessary antibiotic use. Thus, routine administration of antibiotics to patients with false-positive results for CDI has significant negative consequences for patients and hospitals.

The views expressed in this feature do not necessarily represent the views of the journal or of ASM.