NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.
Butler M, Bliss D, Drekonja D, et al. Effectiveness of Early Diagnosis, Prevention, and Treatment of Clostridium difficile Infection [Internet]. Rockville (MD): Agency for Healthcare Research and Quality (US); 2011 Dec. (Comparative Effectiveness Reviews, No. 31.)
This publication is provided for historical reference only and the information may be out of date.
Effectiveness of Early Diagnosis, Prevention, and Treatment of Clostridium difficile Infection [Internet].
Show detailsThe general search identified 1,078 citations from MEDLINE. Of these, 356 studies were pulled for full text screening. Of these 356 references, we included 69 randomized controlled trials (RCTs), systematic reviews, observational studies, and an additional 22 articles obtained from hand searching and review article bibliographies. We excluded 998 articles. A supplemental search for diagnostics identified 519 citations from MEDLINE, of which 516 references were excluded. Figure 4 provides a literature flow diagram. A bibliography of the excluded articles, and their reasons for exclusion, is provided in Appendix D.
Figure 4
Reference flow diagram. KQ = Key Question
Key Question 1. How do Different Methods for Detection of Toxigenic C. difficile Compare in Their Sensitivity and Specificity?
Search Results
We included 13 references that provided comparative data about diagnostic tests of interest. The studies were published from 2001 to 2010. Five studies were from the United States, two were from the United Kingdom and Spain, and one each were from Belgium, Ireland, Israel, and the Netherlands. Table 1 provides a summary of the available comparisons. Overall, these reports included data on seven named immunoassays for toxins A and B, one two-stage method where an immunoassay for glutamate dehydrogenase was combined with an immunoassay for toxins A and B, and two tests to detect gene fragments involved in the production of toxin B. Only three comparative studies included one of the recently FDA-approved toxin gene detection tests. Thus, the number and type of paired (within study) comparisons available for each diagnostic test varied considerably, and not all possible comparisons were available. Evidence summary tables, including study quality items, are available in Appendix C of this report (see Appendix Table C1).
Table 1
Summary of diagnostic comparisons in included studies.
Key Points
- Sixteen paired comparisons of seven immunoassays for toxins A and B provided low-grade evidence that the test sensitivities do not differ. There was moderate-grade evidence for no differences in test specificities for three comparisons and for a difference of 2 percent in one comparison. Otherwise, there was only low-grade evidence for or against differences in test specificities. There was insufficient evidence of differences between all tests that were not directly compared.
- Nine comparisons of two different gene detection tests to toxin immunoassays provided only low-grade evidence to support the notion that the gene-based tests are substantially more sensitive than immunoassays. There was moderate evidence that the test specificities in one comparison did not differ. Otherwise, there was only low-grade evidence for differences in either direction between test specificities. There was insufficient evidence of differences between all tests that were not directly compared.
- There was insufficient evidence to determine whether any differences in sensitivity or specificity between diagnostic tests depend on patient or specimen characteristics or the strain of toxigenic Clostridium difficile.
Quality of the Comparative Studies
All studies used stool specimens from mostly inpatients that were submitted by clinicians to test for Clostridium difficile infection (CDI). However, the clinical scenarios that prompted the clinicians to test for CDI, such as the nature of the patient's diarrhea, or exposure to antibiotics, were not described in many reports. Seven of the 13 studies that provided data mentioned that the stool samples were liquid, unformed, or diarrhea, whereas the other reports did not clearly describe the consistency of the stool specimens. Six of the studies included more than one specimen from some patients, and three studies only reported the total number of stool specimens and not the number of patients. Two studies selected stool samples based on previous diagnostic test results to enhance the percentage of positive tests in their sample, and two included a facility with a recent outbreak of CDI or high prevalence. Thus, the reviewed reports were somewhat deficient in reporting pertinent information about patient selection criteria and the spectrum of patients/specimens included the comparisons (Appendix Table C2).
Differences within studies in the timing and handling of specimens for the different tests being compared were not a major issue in the reviewed studies. Verification using the reference standard was applied consistently to all stool specimens. However, the same reference standard was not used in all studies. Five of the 13 studies used a cell cytotoxicity test as the reference, five used a cell cytotoxicity test in conjunction with toxigenic culture, one used a toxin immunoassay in conjunction with toxigenic culture, one used multiple immunoassays for toxins A and B in conjunction with toxigenic culture, and one used an in-house gene detection test. None of the reference methods that were used are a true gold standard in that they are not 100 percent sensitive or specific for toxigenic C. difficile and their accuracies are not all the same. Within each study, the diagnostic tests were carried out independently of each other although the reports usually did not state that each test was interpreted without knowledge of other results. Only two reports explicitly stated that all diagnostic tests being compared, including the reference test, were conducted in a blinded manner. Sometimes the independence of the tests could be inferred from their sequence and the time needed to get results.
The handling of indeterminate test results presents problems when comparing the sensitivity and specificity of diagnostic tests. Some investigators repeated indeterminate tests and used the result of the second test as recommended, although some repeated tests were also indeterminate. Some assumed indeterminate results were negative and thereby could have inflated the number of false negatives. Some comparisons excluded indeterminate results; thus, the varying number of indeterminate tests did not count for or against a test. However, differences in the number of indeterminate results produced by different tests resulted in some differences in the stool specimens being used to compare the tests. Other types of subject or specimen withdrawal were not an issue in the studies that were reviewed.
Detailed Analysis
Comparisons of Immunoassays for Toxins A and B
As summarized in Table 2, none of the seven immunoassays for toxins A and B was compared to all others. When more than one study compared the same two immunoassays, the heterogeneity in the differences in sensitivity was significant in only one out of nine cases. None of the nine pooled comparisons based on two to four studies indicated that any of the immunoassays were more sensitive than another. The pooled estimates of the differences (99 percent confidence interval [CI]) in test sensitivities were 0±6 percent, 1±7 percent, 3±6 percent, 3±7 percent, -1±10 percent, 3±8 percent, 6±12 percent, 1±9 percent, and 3±24 percent. The confidence intervals for single-study estimates of differences in sensitivity were wide. Thus, the available data often could not rule out substantial differences in sensitivities.
Table 2
Comparisons of immunoassays for toxins A and B.
There was some significant heterogeneity in the corresponding estimates of differences in false positives (1 minus specificity) for two of the nine multiple study comparisons of immunoassays for toxins A and B. Ignoring the heterogeneity, the differences (99 percent CI) in false positives were 0±2 percent, 0±1 percent, 2±1 percent, 0±1 percent, -3±3 percent, -1±10 percent, -6±14 percent, 3±2 percent, and 2±2 percent. Thus, the available data often ruled out differences in false positives of only a few percent. One study that compared several immunoassays found some differences in the false positives of approximately 6 percent.
Gene Detection Tests Versus Immunoassays for Toxins A and B
As summarized in Table 3, two studies compared the same tests to detect genes related to toxin B production to the same immunoassay for toxins A and B.32,37 There was significant heterogeneity between the estimated differences in sensitivities for both comparisons; however, in each case both studies suggested the gene-based test was more sensitive than the immunoassay. The pooled estimate of the difference in sensitivities was 17 percent in favor of the gene based test with a 99 percent confidence interval of from 3 to 37 percent in one comparison, and 25 with a 99 percent confidence interval of from -36 to 86 percent in the other comparison. There was no heterogeneity in the corresponding estimated differences in false positive percentages of these tests. The pooled estimate of the differences in the false positives were 0 percent with a 99 percent confidence interval of from 1 percent to 1 percent for one comparison, and 2 with a 99 percent confidence interval of from -1 percent to 5 percent for the other comparison. The percentage of false positives tended to be greater with the gene detection test in the later comparison.
Table 3
Toxin gene detection tests compared to immunoassays.
Three studies provided one pairwise comparisons of a gene detection test to an immunoassay for toxins A and B.32 The sensitivity of the gene detection test was consistently better, although the point difference ranged widely from 3 percent to 56 percent, and the confidence intervals didn't always exclude a difference of zero. The false positives for the gene-based test were approximately 3 percent greater compared to one of the immunoassays for toxins A and B.
The sensitivities of the two gene detection tests in the three studies ranged from 89 percent to 100 percent. In contrast, the sensitivities of the immunoassays for toxins A and B were much more variable, ranging from 44 percent to 86 percent. The methodological differences between studies, including use of different reference tests, might have affected the toxin immunoassays more than the gene detection tests. The estimated sensitivities of the immunoassays were remarkably low (only 44 or 58 percent) in two studies that used the generally most sensitive reference test (toxigenic culture).
Key Question 2. What are Effective Prevention Strategies?
Search Results
We found 1 Cochrane review,41 4 studies on antibiotic prescribing restrictions,42-45 12 on single preventive practices aimed at transmission interruption,46-55,67 and 10 that bundled multiple practices into a prevention strategy.56-65 Only two trials were controlled trials;46,49one was an interrupted time series study,42,66and the remaining studies were before/after designs.43-45,47,48,50-55,66,67 The included studies are provided in Table 4.
Table 4
Prevention interventions.
Eight studies examining risk factors met the inclusion criteria and updated the period following a systematic review20 (Appendix Table C3). Five studies were conducted in the United States,111,166-169 two in Israel,112,170and one in the United Kingdom.171 The average CDI patient sample was 86 patients, with a range of 28 to 154. Studies varied in the degree to which the investigators verified that positive tests reflected disease.
Key Points
- Overall, the evidence available to link prevention strategies to clinically important outcomes, such as CDI incidence, is of low strength and is not extensive.
- Four observations studies and one Cochrane review found prescribing practice interventions decreasing the use of high-risk antimicrobials are associated with decreased CDI incidence. Prescribing practices were also used in multicomponent interventions credited with reducing CDI incidence; however, it is difficult to isolate the specific effects of the prescribing practices.
- One controlled trial found glove use significantly reduced CDI incidence.
- Three observational studies, including two controlled, found disposable thermometer use is likely to reduce CDI incidence.
- No study examined the effect of handwashing, rather than alcohol gels, on CDI incidence. Four observational studies found use of alcohol gels as interventions for other infectious diseases, presumably in the presence of protocols requiring handwashing in the presence of CDI or visible soiling, did not increase CDI incidence.
- Three studies provide low evidence that disinfection with a chemical compound that kills C. difficile spores in the hospital environment prevents CDI, at least in epidemic or hyperendemic settings. Seven studies included disinfection in multicomponent interventions. Disinfection agents examined included hypochlorite solution, hydrogen peroxide, aldehydes, and detergent.
- Ten time series/before-after studies have examined bundled multiple interventions using before/after study designs. Data are insufficient to draw conclusions.
- Risk factors for developing CDI include antibiotic use, substantial chronic illness, hospitalization in an ICU, age, and acid suppression therapy.
- No data on patient harms or harms to hospital staff due to preventive interventions were reported.
- No studies assessed the sustainability of a prevention program beyond an intervention period.
Quality of the Studies
Overall, the quality of the evaluated studies was considered low (Table 4). In the Cochrane review41 focusing on improving antibiotic prescribing practices, the evidence from one article172 was judged to be of “good” quality, and evidence from the others was considered “weak.” The evidence for the 10 single preventive practices aimed at transmission interruption was low because they predominately used before/after design and were done in response to epidemic or hyperendemic conditions. In particular, there is insufficient evidence that handwashing is associated with reduced CDI incidence, as no study assessed this intervention. Of the four studies assessing alcohol based rubs or gels, only one had concurrent controls. Thirteen studies examining environmental disinfections were all before/after studies, generally done in response to epidemics.
For the 10 articles that described multiple component preventive interventions, none had concurrent controls or was blinded, and there was considerable variability in the types of interventions, so pooling could not be done. In addition, it was indeterminable to attribute decreases in CDI incidence to any single intervention in all of these studies.
Detailed Analysis
Due to the low-quality studies, we provide a qualitative narrative of the evidence for prevention practice interventions.
Antibiotic Use
The five studies summarized in the Cochrane review,41 and the additional four individual studies here,42-45 found that changes in antimicrobial education, policies, or formularies, which result in decreasing use of high-risk antimicrobials, are associated with decreased CDI incidence. It was not possible to clearly isolate the impact of the antibiotic-related interventions in the studies examining multiple interventions.58-60,62,63 In the individual studies, which were usually done in response to outbreaks, interventions in addition to those aimed at antibiotic use may have been done but not reported. The interventions and antibiotics targeted for reduction differed among the various studies.
The Cochrane review41 determined the impact of interventions to improve antibiotic prescribing practices for hospital inpatients on CDI incidence. The authors found that four interventions were associated with significant reductions in CDI incidence62,172-174 and that one was associated with a nonsignificant trend toward a reduction.61
A prospective controlled interrupted time series42 of an antibiotic improvement intervention on three acute medical wards for elderly people with 21-month predefined pre- and postintervention periods, evaluated a “narrow-spectrum” antibiotic policy (reinforced by an established program of audit and feedback of antibiotic usage and CDI rates). The program targeted broad-spectrum antibiotics (cephalosporins and amoxicillin/clavulanate) for reduction and narrow-spectrum antibiotics (benzyl penicillin, amoxicillin. and trimethoprim) for increase. CDI rates decreased significantly with incidence rate ratios of 0.35 (95 percent CI 0.17 – 0.73). Incidence of Methicillin-resistant Staphylococcus aureus (MRSA), the control, did not change significantly.
The effect of a new antibiotic policy favoring piperacillin-tazobactam over cefotaxime on the long-term incidence of CDI and antibiotic utilization in a large elderly medicine unit was studied in a before/after observational study.43 Restrictions were associated with reduced cefotaxime use and reduced CDI incidence. Subsequently, the piperacillin-tazobactam became unavailable at the end of 2001. Cefotaxime use and CDI incidence rates increased during 2002.
In a geriatrics department of a university hospital, antimicrobial recommendations for treatment of several common infectious diseases were changed from broad-spectrum cephalosporins to other drugs thought to be less likely to induce CDI.44 Investigators changed department policy to reflect these recommendations, educated providers, monitored antibiotic use, and gave periodic feedback to providers. Cephalosporin use dropped, and the relative risk of CDI decreased to 0.31 (95 percent CI 0.93 to 0.10) compared with usage before the policy change.
In a geriatrics department of another university hospital, broad-spectrum cephalosporin use was restricted due to an increase in CDI incidence.45 In the following year, cephalosporin use decreased 92 percent, and CDI incidence decreased 50 percent from the previous year incidence. CDI incidence did not change in other hospital departments.
Measures to Reduce Transmission
Gloves
One controlled trial examined the use of gloves to prevent C. difficile transmission, with CDI incidence monitored by active surveillance.46 An intensive education campaign on two wards urged personnel to use gloves when handling body substances, and gloves were made easily available to personnel working with patients. Two other wards with no education campaign served as control wards, and gloves on these wards were stocked in supply rooms. Incidence of CDI decreased significantly from 7.7 cases/1,000 patient discharges during the 6 months before intervention to 1.5/1,000 during the six months of intervention on the intervention wards. No significant change in CDI incidence was observed on the control wards. Asymptomatic C. difficile carriage also decreased significantly on the intervention wards but not on the control wards. The cost of 61,500 gloves (4,505 gloves/100 patients) used was $2,768 for the glove-using wards, compared with $1,895 (42,100 gloves; 3,532 gloves/100 patients) on the control wards.
Disposable Thermometers
Three studies, one randomized crossover design,49 and two before/after studies without concurrent controls47,48 have shown that use of disposable thermometers prevent CDI. In one hospital with an increased CDI incidence, 21 percent of electronic rectal thermometer handles were contaminated with C. difficile.47 Efforts to reinforce infection control practices were already in place, but CDI incidence remained elevated. A before/after trial was conducted in that hospital and a chronic care facility to determine if use of disposable thermometers instead of multiple-use electronic rectal thermometers would reduce the CDI incidence. Surveillance for CDI was active, but toxin was detected with a latex agglutination test. During the 6-month postintervention period, the CDI incidence decreased from 2.71/1,000 patient days to 1.76/1,000 patient days in the acute hospital and from 0.41/1,000 patient days to 0.11/1,000 patient days in the skilled nursing facility. The harms associated with use of disposable thermometers were costs for purchase of disposable thermometers and the need to dispose of these thermometers. In these institutions, annual outlays increased from $7,731 to $14,055. These costs were offset by the need to purchase fewer electronic thermometers and to sterilize them periodically and by decreased costs of treating CDI cases.
In a later report, the same group reported that the rate of C. difficile infections increased from 1991 to 1993, although it was unclear how many patients had symptoms of disease with C. difficile.48 One ward used disposable tympanic membrane thermometers instead of disposal oral or rectal thermometers. Different interventions were implemented in two other wards. Regression analysis determined that the C. difficile infection rate decreased 40 percent (relative risk [RR], 0.59, 95 percent CI, 0.47-0.67).
A randomized, controlled crossover study compared the use of disposable thermometers with electronic thermometers to prevent nosocomial CDI.49 Twenty hospital wards were randomly assigned to disposable thermometers or electronic thermometers for 6 months, and then the assignments were reversed for 5 months. CDI rates were reduced 44 percent (P=0.026, 95 percent CI, 0.21 to 0.93) with disposable thermometers compared to electronic thermometers. Rates of nosocomial diarrhea or nosocomial infections did not differ significantly between the two groups. A cost analysis estimated that the hospital using disposable thermometers would need to spend an additional $5,926 to prevent a single CDI case. It was estimated that a CDI case resulted in $2,000 to $6,000 in excess costs.
Handwashing
No study addressed whether handwashing was associated with reduced CDI incidence. Many institutions encourage the use of alcohol-based rubs or gels for hand hygiene unless hands are grossly soiled or unless a health care worker has had potential contact with C. difficile either from patient contact or environmental contamination. Neither alcohol nor soap will kill C. difficile spores, but when health care workers wash hands properly with soap, most spores are removed because of friction and the detergent action of soap. Complicated recommendations are difficult to remember and implement, and one concern has been that health care workers will use alcohol-based rubs or gels in circumstances where handwashing is preferred.
Four studies have addressed this concern. One 2-year, prospective, controlled, crossover trial compared alcohol-based hand gel provided in addition to hand soap containing the antimicrobial 0.3 percent chloroxylenol with antimicrobial soap alone in two intensive care units.50 In units using adjuvant alcohol-based gel, there was a significant, sustained improvement in the rate of hand hygiene adherence but no detectable change in the incidence of healthcare-associated CDI (diagnosis determined by clinicians).50 Employees still had access to soap and water when their hands were soiled or when they were caring for a patient with C. difficile, and if workers used soap and water in these circumstances, it would have decreased the likelihood that differences in CDI rates would be detected.
The second study used a before/after design.51 Hospital employees were encouraged to wash hands with the antimicrobial 0.3 percent triclosan in the first 3-year period, and an alcohol-based hand rub with 62.5 percent ethyl was placed in dispensers in inpatient and outpatient clinic rooms in the next 3 years. There was a 21 percent decrease in new, nosocomially acquired MRSA isolates and a 41 percent decrease in vancomycin-resistant enterococci (VRE) isolates, but the incidence of new CDI cases remained similar (diagnosis determined by clinicians/toxin A assay).51
A retrospective time-series analysis, the secondary objective, was done to determine the relationship between use of alcohol-based hand rub and antibiotic consumption on the incidence of CDI.52 CDI incidence was determined retrospectively from records of patients put in isolation for CDI. Multivariable time series analyses showed no association between alcohol-based hand rub and CDI incidence. Macrolide and third-generation cephalosporin use was associated with increased CDI incidence after lag times of 1 to 3 months.
A retrospective, interventional time-series analysis was used to determine the effects of two interventions on CDI incidence.66 The interventions were promotional campaigns to encourage use of alcohol-based hand rub for hand hygiene. Time series analysis was done with autoregressive integrated moving average models. There was no association between alcohol-based hand rub and CDI incidence.
Disinfection
Four studies examined if disinfection reduces the incidence of CDI as a single component intervention,53-55,67 and seven studies included disinfection in multicomponent interventions.56,57,59,60,63,65,68 Disinfection agents examined included hypochlorite solution, hydrogen peroxide, aldehydes, and detergent.
Three studies examined hypochlorite solution as a single intervention. One before/after intervention investigated whether cleaning patient rooms that tested positive for C. difficile toxin with unbuffered with 1:10 hypochlorite solution reduced the incidence of CDI in three patients' units.53 Before the intervention, patient rooms were cleaned with quaternary ammonium. In one housing bone marrow transplant patients and having the greatest rate before the intervention, the CDI incidence rate decreased significantly, from 8.6 to 3.3 cases per 1,000 patient-days (hazard ratio 0.37, 95 percent CI, 0.19 to 0.74) after hypochlorite was used to clean rooms. In the other two with lesser rates before the intervention, there was no significant change. In response to a subsequent outbreak of VRE infections, the hospital used quaternary ammonium solution for all patient room disinfection. The incidence of VRE infection decreased, but the CDI incidence rate increased. Hypochlorite disinfection was reinstituted and the CDI incidence rate subsequently decreased. A followup report documented subsequent increases in incidence and further interventions to control CDI.
An epidemiological investigation of an outbreak of CDI occurring in a single ward of a Michigan hospital documented nosocomial acquisition from the environment.54 After use of unbuffered hypochlorite to disinfect wards, contamination decreased and the outbreak ended. Subsequently, it was shown that phosphate-buffered hypochlorite was even more effective for disinfection.
Hypochlorite was used in various ways in conjunction with other interventions to prevent CDI in seven studies (multiple intervention table part B).56,57,59,60,63,65,68 The effect of the hypochlorite disinfection cannot be isolated from the other intervention components.
A high rate of CDI was noted in three hospitals joined in a single health care system. Hospitals changed the disinfectant used for the discharge cleaning of rooms of patients with CDI from a quaternary ammonium compound to dilute bleach.67 There was a 48 percent reduction in the prevalence of C difficile after the bleaching intervention (P=0.0001, 95 percent CI, 36 to 58).
Two before/after studies were conducted to evaluate whether disinfection with hydrogen peroxide as part of multiple component interventions reduces CDI incidence.57,63 In the first study, an abrupt increase in nosocomial CDI (defined as diarrhea with a positive toxin test) incidence led to multiple interventions in attempts to control the outbreak. Surveillance was based on laboratory and patient medical records.57 A liquid vapor hydrogen peroxide decontamination system was used to decontaminate five high incidence wards of C. difficile organisms.57 There followed a slight decrease in nosocomial CDI incidence. Liquid vapor hydrogen peroxide was then used to decontaminate patient rooms vacated by patients with CDI throughout the hospital on an ongoing basis. Nosocomial CDI incidence continued to decrease and remained at levels roughly equivalent to rates prior to the outbreak. Quality of the diagnosis and surveillance system was good. No harms to hospital personnel, patients, or equipment were observed. The authors noted that the area to be decontaminated must be appropriately sealed, hydrogen peroxide levels outside the area being decontaminated must be closely monitored, and hydrogen peroxide concentrations within the decontaminated area must be reduced to less than 1 part per million before allowing patients or health care workers to re-enter. A subsequent study by the same investigators reported that hydrogen peroxide vapor disinfection was feasible in their hospital.175 The peroxide vapor disinfection took 2 hours and 20 minutes to complete compared with 32 minutes for routine cleaning. The median cumulative times for all phases of cleaning and disinfection were 234 minutes (range 174–838) for peroxide vapor compared with 55 minutes (range 28–256) for conventional hypochlorite.
In the second study, 7 percent accelerated hydrogen peroxide was used for terminal disinfection of rooms of patients with CDI and comprehensive ward disinfection with sodium hypochlorite was done when three or more nosocomial CDI cases (defined as cases with positive toxin or with endoscopic or histological evidence of pseudomembranous colitis) remained elevated.63 Within 4 months of the time infection prevention measures were implemented, the investigators also took several steps to reduce antibiotic use. Nosocomial CDI incidence fell abruptly within 1 month of the changes in antibiotic use.
In one study using aldehydes as part of a multiple-component intervention, a cluster of CDI in a surgical ward led to a hospitalwide surveillance and control program.55 Control interventions included terminal room disinfection with 0.04 percent formaldehyde and 0.03 percent glutaraldehyde in wards with a cluster of two or more nosocomial CDI cases per month. During a 12-month period, the quarterly incidence of nosocomial CDI remained unchanged. C. difficile spores were recovered from 36.7 percent of the surfaces of case patient rooms versus 6.7 percent in control rooms. Subsequently, more intensive control measures were evaluated, which included daily meticulous room disinfection for each sporadic nosocomial CDI case. Surface disinfection reduced the contamination level fourfold (p = 0.04). In the following 12 months, the nosocomial CDI incidence fell to 0.3/1,000 admission (protective efficacy 73 percent, 95 percent CI, 46–87 percent). Multiple interventions, including disinfection, were used to control the outbreak. The study provides low evidence that disinfection, in this case with aldehydes, might have had a role in terminating the outbreak.
These ten studies provide low evidence that disinfection with a chemical compound that kills C. difficile spores in the hospital environment prevents CDI, at least in epidemic or hyperendemic settings. Decreased CDI incidence might have been from natural variation (regression to the mean) in some or all studies. As stated previously, disinfection was one of multiple interventions used to prevent CDI in seven studies; it is difficult to impossible to know which intervention or combination of interventions might have led to reduced CDI incidence.
Multiple Component Studies
Ten studies described the use of multiple preventive measures to control epidemic CDI, or endemic CDI that was felt to be excessive. Tables 5 and 6 list the categories of interventions in each of these articles. The number of interventions and the specific nature of any particular interventions varied widely. Studies employed between two and nine different types of interventions, including steps to optimize antimicrobial (six studies),58-63 enhanced surveillance (two studies),59,60 intensified staff education about infection prevention (three studies),60,62,63 new isolation procedures (four studies),59,61,63,64 and “enteric precautions” (two studies).58,61 Two studies emphasized handwashing61,64 and one alcohol-based gel for hand disinfection.60 Health care workers were required to wear gloves in three studies,61,62,64 and use of gowns for patient contact was required in two studies.61,64 Visitors were asked to comply with infection prevention procedures in one study.64 New dedicated patient care equipment was purchased in two studies,60,63 and in one of these, cleaning of dedicated patient equipment was intensified. Disposable rectal thermometers were used in one study.63 Intensified environmental cleaning was implemented in six studies.59-64 CDI patient movement was restricted in two studies.59,64
Table 5
(A) Studies of multiple interventions used together to reduce CDI incidence.
Table 6
(B) Studies of multiple interventions used together to reduce CDI incidence.
Investigators often placed greater weight on one intervention over others because the timing of decreased CDI incidence appeared to follow implementation of a particular intervention. However, the time it takes for many interventions to become adopted in health care settings and the variance expected in disease incidence led us to conclude that it was not possible to attribute decreases in CDI incidence to a single intervention in any of these studies. Natural fluctuations are such that all outbreaks diminish after variable periods of time so that assigning causality to individual or a collection of prevention measures is impossible. The evidence from these studies that any single intervention or combination of interventions prevents CDI was low.
Harms
Harms, beyond cost, were not addressed in any study.
Risk Factors
Identified CDI risk factors can provide clues to researchers and health care providers for where to target prevention strategies. We identified one systematic review reviewed CDI risk factor literature through 199720 and 12 risk factor studies published after the review. Bignardi's systematic review identified risk factors with “substantive” evidence: age, severity of underlying diseases, nonsurgical GI procedures, nasogastric tube, acid suppression medications, ICU, length of stay, duration of antibiotic course, and multiple antibiotics.20 Five studies identified specific antibiotics or antibiotic classes with increased CDI risk111,168,169,176,177 (Table 7), and two studies found that antibiotic use in general was associated with increased risk for CDI.111,171 Consistent with Bignardi's findings, the more recent literature also identified severe underlying disease as a risk factor in four studies112,166,167,171 and acid suppression in one.112
Table 7
Summary of risk factors for CDI.
Sustainability
No studies addressed the sustainability of a prevention program.
Key Question 3. What are the Comparative Effectiveness and Harms of Different Antibiotic Treatments?
Search Results
Eleven randomized clinical trials were identified that evaluated different antimicrobials (or different doses of a single drug) available for treatment of CDI in the United States. These 11 studies, published from 1978 to 2009, ranged in size from 39 to 629 subjects. Table 8 provides a breakdown of the trial comparators. Vancomycin is the most frequently studied antimicrobial, examined in 8 of the 10 studies. The most frequent comparison was vancomycin versus metronidazole (three studies, one of which also included fusidic acid and teicoplanin treatment arms, which are not included in this analysis), followed by two studies of vancomycin versus bacitracin. The remaining comparisons (vancomycin vs. nitazoxanide, vancomycin vs. fidaxomicin, vancomycin high dose vs. low dose, vancomycin vs. placebo, metronidazole vs. nitazoxanide, and metronidazole vs. metronidazole plus rifampin) all occurred in single studies. Treatment duration was 10 days in 9 of 11 studies, with the other two having durations of 7 and 5 days. The typical study followup period was 21 to 31 days. The largest patient sample was 629; most studies were in the range of approximately 40 to 60 patients. (See Appendix Table C4.) Two studies that did not meet inclusion criteria merit brief mention: one179 appears to report on the same subjects included in another publication,78 while another180 has been presented in abstract form only.
Table 8
Summary of trial comparators for 10 trials of antibiotic treatment of CDI.
Key Points
- Overall, study quality is low.
- Vancomycin and metronidazole, the most frequently clinically used antimicrobials, were the most frequently compared antimicrobials.
- Three RCT comparisons of vancomycin to metronidazole, with a total of 335 pooled subjects, found no significant differences in any examined outcome.
- One RCT comparing vancomycin to metronidazole, using a prespecified subgroup analysis of 69 patients, found a small but significant increase in the proportion of subjects with severe CDI who achieved initial clinical cure with vancomycin, using a treatment-received analysis. This difference was not significant using a strict intention-to-treat analysis.
- One study demonstrated that recurrence was significantly decreased with fidaxomicin versus vancomycin; initial cure was not significantly different between fidaxomicin and vancomycin.
- The decrease in recurrence seen with fidaxomicin use appeared to be limited to those patients with non-NAP1 strains.
- Harms were not reported with sufficient detail to compare the risks of any particular antimicrobial with another antimicrobial.
- When harms were reported, they were generally not serious (nausea, emesis, etc.) and transient.
Minor Key Points
- No other head-to-head trial demonstrated superiority of any single antimicrobial for initial clinical cure, clinical recurrence, or mean days to resolution of diarrhea.
- Combination therapy with rifampin and metronidazole resulted in significantly higher mortality when compared to treatment with metronidazole only.
- Pooled data of 104 subjects comparing vancomycin to bacitracin showed significantly higher rates of organism or toxin clearance for vancomycin.
- No data were available to assess the importance of general patient characteristics or the strain of organism on the effectiveness of an antimicrobial.
Quality of the Studies
Overall study quality is low. Only two studies specified that the investigators (who also assessed outcomes) were blinded with respect to treatment.70,82 Quality summary tables are available in Appendix C of this report (see Appendix Tables C5 and C6). Strength of evidence is summarized in Appendix Tables C7 and C8.
Detailed Analysis
As vancomycin and metronidazole are the most frequently employed antimicrobials, and therefore of greatest interest to clinicians, results are broken into two sets: (1) vancomycin versus metronidazole and (2) all other comparisons of standard treatment trials.
Initial Cure
The percentage of subjects initially cured with vancomycin ranged from 84 percent to 94 percent among individual studies, with a mean value of 88 percent (Table 9). For subjects treated with metronidazole, the individual cure rates ranged from 73 percent to 94 percent, with a mean value of 81 percent. The relative risk for initial cure comparing vancomycin to metronidazole was 1.08 (95 percent CI 0.99 to 1.19).
Table 9
Initial clinical cure (# subjects / # randomized) for vancomycin versus metronidazole.
With the exception of vancomycin versus placebo, no other treatment comparison resulted in significant differences in initial clinical cure (Table 10).
Table 10
Initial clinical cure (# subjects / # randomized) for all other standard treatment trials.
Clinical Recurrence
The percentage of subjects meeting the investigator-determined definition of recurrent disease (after meeting criteria for initial cure) ranged from 7 percent to 17 percent with vancomycin, with a mean value of 11 percent. For metronidazole the range was 5 percent to 21 percent, with a mean value of 12 percent. (Table 11) The relative risk for recurrence after vancomycin treatment compared to metronidazole was 0.92 (95 percent CI, 0.47 to 1.77)
Table 11
Clinical recurrence: # subjects / # initially cured (percent) for vancomycin versus metronidazole.
Only the comparison between fidaxomicin and vancomycin showed a statistically significant difference (15 percent vs. 25 percent, P = 0.005); in all other trials there was no significant difference in percentage of patients with recurrence. Between trial comparisons for the percentage of patients with recurrence are of uncertain relevance because of the variable definitions of recurrence and duration of followup. (Table 12).
Table 12
Clinical recurrence: # subjects / # initially cured (percent) for all other standard treatment trials.
Mean Days to Resolution of Diarrhea
Two of the three vancomycin versus metronidazole studies reported the mean time to resolution of diarrhea.73,76 No differences were seen between treatment arms (Table 13).
Table 13
Mean days to resolution of diarrhea/clinical improvement for vancomycin versus metronidazole.
No other treatment comparison resulted in significant differences in mean days to resolution of diarrhea (Table 14).
Table 14
Mean days to resolution of diarrhea/clinical improvement for all other standard treatment trials.
All-Cause Mortality
Mortality was rare overall, in part due to the short study-followup periods. There were five deaths in each arm among the 335 subjects enrolled in studies comparing vancomycin with metronidazole (Table 15). Wenisch73 evaluated four drugs, including two not evaluated in this review, but did not provide mortality data by subject. Depending on in which study arm the mortalities occurred in the Wenisch study,73 there were between 10 and 13 total deaths in studies comparing vancomycin to metronidazole. Even if all three deaths in this study occurred in one arm, the difference in mortality could not reach statistical significance.
Table 15
All-cause mortality (# subjects / # randomized) for vancomycin versus metronidazole.
All-cause mortality was significantly higher for combination metronidazole plus rifampin versus metronidazole alone (32 percent versus 5 percent).71 There were no differences in all-cause mortality in any of the other treatment comparisons (Table 16).
Table 16
All-cause mortality (# subjects / # randomized) for all other standard treatment trials.
Other Outcomes
Where the outcomes were reported, no differences were found between vancomycin and metronidazole for clearance of toxin,73 laboratory-confirmed relapse,73 or persistence of the organism76 (Table 17). The clinical relevance of these outcomes is uncertain.
Table 17
Other outcomes (# subjects / # assessed) for vancomycin versus metronidazole.
Pooled data of 104 subjects comparing vancomycin to bacitracin showed significantly higher rates of organism or toxin clearance for vancomycin.74,77 No other differences were found in reported outcomes (Table 18).
Table 18
Outcomes for all standard treatment trials.
Harms
Reported adverse events were relatively uncommon, minor, and not associated with one drug compared with the other. One study reported two episodes of intolerance (nausea and vomiting) leading to subject withdrawal, one in each treatment arm.76 Another reported a subject with emesis that developed while on metronidazole, which resolved when treatment was changed to vancomycin; in the same study, another subject developed nausea while on vancomycin, which resolved when treatment was changed to metronidazole.70 The third study reported “gastrointestinal discomfort” (which did not result in cessation of therapy) in 10 percent of subjects receiving metronidazole, compared to none with vancomycin, a difference that did not reach significance.
Disease Severity
Only one study stratified patients by disease severity at the time of screening.70 Severity was dichotomized into two outcomes: mild or severe disease. This trial stratified treatment based on disease severity (mild versus severe). Sixty-nine subjects, 31 who received vancomycin and 38 who received metronidazole, met the prespecified definition of severe disease. Patients with two or more of the following were considered to be severe: 60 years old or older, temperature above 38.3 degrees Celsius, albumin level less than 2.5 mg/dL, or peripheral white blood count greater than 15,000 cells/mm3 within 48 hours. Using a treatment-received analysis, the authors reported that initial cure was more common among those receiving vancomycin (97 percent versus 76 percent), with a relative risk for initial cure of 1.27 (95 percent CI, 1.05 to 1.53). In a subsequent response181 to several letters,182-184 they reported a revised result, which incorporated a modified intention-to-treat analysis (including subjects who died in the first 5 days of therapy), and reclassification of two subjects as being initially cured. This slightly changed the relative risk for initial cure to 1.28 (95 percent CI, 1.03 to 1.59). However, using a strict intention-to-treat analysis, which includes subjects intolerant of therapy, lost to followup, and early deaths, and the original classification of initial cure, the percentage cured with vancomycin versus metronidazole was 79 percent versus 66 percent. This corresponds to a relative risk for initial cure of 1.20 (95 percent CI, 0.92 to 1.57) (Table 9). This is minimally changed to 1.20 (95 percent CI, 0.93 to 1.54) if the two subjects initially classified as failures are reclassified as cures. No other significant differences in outcomes were found by disease severity.
C. difficile Strain
A single study assessed initial cure and recurrence by strain, categorized as North American pulsed-field gel electrophoresis type 1 (NAP1) versus non-NAP1.79 Strain data was available for 324 of the 629 (51.5%) participants. For initial cure, no significant difference was observed, regardless of strain. However, among patients with non-NAP1 strains, those treated with fidaxomicin recurred less frequently than those treated with vancomycin (10 percent versus 28 percent; P < 0.001), whereas among patients with the NAP1 strain recurrence was similarly frequent regardless of treatment.
Patient Characteristics
Our search did not identify any evidence for comparative effectiveness by general patient characteristics such as age, gender, or treatment setting.
Resistance of Other Pathogens
The impact of treatment for CDI on other pathogens has not been addressed by the available studies that directly assigned subjects to different drugs. From observational studies, there is some evidence that treatment with either metronidazole or vancomycin can cause an increase in the incidence in the carriage of vancomycin resistant enterococci185,186 however, the magnitude of this effect and the clinical significance are uncertain.
Key Question 4. What are the Effectiveness and Harms of Nonstandard Adjunctive Interventions?
Search Results
A total of five RCTs on nonstandard adjunctive treatments of CDI (Table 19) and 13 studies that addressed prevention of CDI (Table 20) formed the basis of this analysis. Four of the studies on treatment of CDI compared a nonstandard intervention with an active control, that is, a standard antibiotic treatment for CDI, oral vancomycin or metronidazole.80-83 One study compared a nonstandard intervention with placebo.84 All of the 13 prevention studies compared the nonstandard intervention to placebo rather than to another intervention, reflecting the current state of this area of science. Five of the 13 prevention studies analyzed antibiotic-acquired diarrhea as a primary outcome and CDI as a secondary outcome.83,85,87,89,188 Numerous published case reports, as well as nonexperimental studies, describe additional nonstandard approaches for treatment of CDI and their possible harms (Table 21).
Table 19
Nonstandard intervention for treatment of initial and recurrent CDI.
Table 20
Probiotic or prebiotic interventions for prevention of initial and recurrent CDI.
Table 21
Summary of case studies/series and potential harms of nonstandard interventions for CDI.
Due to the heterogeneity of the interventions, quantitative analysis was not possible. We therefore provide a narrative review of the literature.
Key Points
- Overall, study quality was low.
- C. difficile immune whey in one study of 38 patients was similar to standard antibiotic treatment with metronidazole in treating recurrent CDI.
- Colestipol plus metronidazole in one study was not more effective than placebo plus metronidazole.
- Administration of a probiotic to treat CDI in critically ill patients increases risk for greater morbidity and mortality from fungemia without any known benefit.
- There is low-strength limited evidence that the nonstandard interventions in this review are not more effective than placebo for primary prevention of CDI.
- There is low-strength limited evidence from one subgroup analysis that a prebiotic may reduce diarrhea recurrence in patients treated for CDI more so than placebo with standard antibiotics.
- There is limited moderate-strength evidence from one study that monoclonal antibodies are effective in preventing recurrence of CDI.
- There is limited low-strength evidence from 6 case studies/series with 60 patients that fecal flora reconstitution is effective in treating recurrent CDI for up to 1 year.
- Data are inconclusive about the benefit of intravenous immunoglobulin as an adjuvant treatment for severe CDI.
- Definitions of CDI with regard to diarrhea, that is, number and consistency of stool, were inconsistent across studies.
Quality of the Studies
The level of the quality of the evidence is low. Several study limitations lowered the quality of their findings. Among the most common were lack of a power analysis, inadequate power to detect significant differences, lack of an intent-to-treat analysis, and failure to define allocation concealment. In one study, the findings of subjects with CDI at the start of a nonstandard intervention were combined with those who developed CDI after the intervention.84 The problem of a nonstandardized, incomplete, or unspecified definition of CDI has already been noted. In one study, a culture of C. difficile (which could have indicated a nontoxigenic strain of the organism) was accepted in place of, or in addition to, a toxin test for the definition of CDI for some patients.84 Longer term followup for CDI incidence or recurrence sometimes relies on reports of diarrhea without retesting for C. difficile toxin. Although probiotics may have been intended solely for prevention of recurrent CDI in some studies, they were included among treatments for recurrent CDI because the probiotic was administered concurrently with a standard antibiotic during treatment and not after recurrent CDI was cured.80,83 Thus, it is not possible to restrict the effect of the probiotic for prevention of future CDI recurrence only. Whether the results of CDI as a secondary outcome are weaker than the primary outcome of AAD due to an underpowered subgroup analysis cannot be determined. There was known lack of adequate power for the primary outcome in one of these studies,87 and no power analysis for the primary outcome was reported in the other four studies.83,85,89,188 There was lack of standardization of the active control in two studies, allowing subjects to receive an antibiotic for CDI as prescribed by their physicians.91,95 Summaries of study quality and strength of evidence are provided in Appendix Tables C9 and C10.
Detailed Analysis
Defining the Outcome of CDI
The operative definition of diarrhea, which is part of the definition of CDI, varied among the studies for prevention and treatment of CDI (Tables 19 and 20). Six of the studies defined diarrhea as three or more loose or liquid stools per day for 2 days.80,81,83-85,95 One study required that same number and consistency of stools but for only 1 day,86 and another study did not require the three stools per day to be loose or liquid.83 One study required two liquid stools on 3 or more days.90 The most liberal definition of diarrhea was one to two loose stools per day.91 Diarrhea due to C. difficile was not explicitly defined in four studies (6 percent).82,87-89
Treatment of CDI
The effectiveness of two types of nonstandard interventions were compared for treating CDI, agents that bind or absorb C. difficile toxins,82,84 and probiotics that aim to recolonize the intestinal flora with nonpathogenic bacteria80,81,83 (Table 19). All interventions were administered orally. Probiotics were the only intervention administered as an adjunct to standard antibiotic treatment for CDI;80,81,83 the other nonstandard interventions were administered independently. The probiotic in two studies contained Sacchromyces boulardii80,83 and in one it contained Lactobacillus plantarum.81
Subjects in the treatment studies had a mean age ranging from 58 to 67 years. Females comprised more than 70 percent of the sample in three of the six studies,80,81,83 and, in one study, the age and gender of subjects were not reported.84 Subjects were hospital inpatients in two studies.80,83,84
The findings of the studies in Table 19 are presented in the same direction, that is, as CDI resolution (versus treatment failure) to facilitate comparison and interpretation. In all studies of CDI treatment, the main outcome was the incidence of resolving CDI, which was defined as diarrhea in patients with a positive stool test for C. difficile toxin.
Treatment of Primary CDI
The rate of resolution of CDI was the lowest in the study comparing an absorptive resin (25 percent of subjects) to placebo (21 percent of subjects); no statistical results were reported.84 Resolution rates for probiotic (81 percent of patients) compared to placebo (76 percent of patients) were not statistically different in another study.83
Treatment of Recurrent CDI
In three comparative treatment studies the subjects recruited were treated for a recurrent (rather than an initial) episode of CDI.80-82 A third study conducted a subanalysis of their subjects with recurrent CDI.83 In all four studies, the nonstandard intervention was probiotic. There was no significant difference in the resolution of CDI between the interventions compared in three of the studies80-82 based on reported statistics or those conducted by the reviewers. In the study that analyzed a subset of their patients with recurrent CDI, a significantly higher percentage of subjects on a standard antibiotic plus a probiotic resolved diarrhea compared to those on a standard antibiotic and a placebo.83
Prevention of Primary CDI
The nonstandard interventions investigated for preventing CDI were (1) probiotics,85-90 (2) a prebiotic (oligofructose) that aims to support a normal ecology of bacteria,91,92 and (3) a monoclonal antibody to C. difficile toxins95 (Table 20). Six different probiotics were tested, and in two of the eight studies, the probiotic contained more than one strain of bacteria.88,90 Seven of the 12 CDI prevention trials using nonstandard interventions focused on primary prevention, i.e., avoiding a first occurrence of CDI.85-91 All of the studies of primary prevention of CDI investigated either a probiotic (six studies) or a prebiotic (one probiotic). Two studies that tested a nonstandard intervention for treating CDI also investigated its ability to prevent CDI recurrence.80,81
Subjects in the primary prevention studies had a mean age of 47 to 77 years. Females comprised less than one-third of the sample in two studies85,89and, in one study, the age and sex of the sample were not reported.88 Subjects in all of the primary prevention studies were hospitalized patients.
The overall incidence of CDI across intervention groups was relatively low, ranging from 2 percent to 9 percent. Only one of seven studies, which investigated a mixture of two probiotics (L. casei and S. thermophilus), showed a significantly lower incidence of CDI diarrhea compared to placebo;90 the investigators of this study acknowledged that the study was underpowered to detect a significant difference greater than by chance. In four studies, statistical testing was not reported.86,87,89,91 Based on reported statistics, or those conducted by the reviewers, there was no significant difference in the recurrence of CDI between any of the interventions and placebo in the six other studies.85-89,91
There is disagreement in the research community regarding the appropriateness of pooling results of probiotics due to the heterogeneity of probiotic organisms used and variability in dosing. We provide a forest plot (Figure 5) of the effects of probiotics on overall incidence of CDI from the primary prevention probiotic trials for those who view such aggregation as reasonable. The pooled RR is 0.40 (95 percent CI, 0.20 to 0.83). The prebiotic trial showed no effect.
Figure 5
Overall incidence of CDI from probiotic primary prevention trials. CI = confidence interval; M-H = Mantel-Haenszel test
Prevention of Recurrence of CDI
Five studies investigated the effectiveness of a nonstandard intervention to prevent the recurrence of CDI (Table 20). Three studies investigated a probiotic,80,81,83 one a prebiotic,92 and one a monoclonal antibody to C. difficile.95 The mean age of subjects ranged from 58 to 75 years. Females comprised 70 percent or more of the subjects in three studies.80,81,83 Hospital inpatients comprised the sample in two studies83,91 and were included along with nonhospitalized subjects in a second study.80 The overall recurrence rate of CDI across intervention groups ranged from 6.5 percent to 34.5 percent.
A significantly lower rate of CDI recurrence was reported in two studies following administration of the prebiotic oligofructose92 or a monoclonal antibody to C. difficile toxins A and B.95 In both studies, the recurrence rate of CDI was approximately three times as great in subjects on placebo compared with the intervention. There was no significant difference in the recurrence of CDI in subjects taking probiotics80,81 compared to controls. In one study comparing a probiotic versus placebo as adjuvants to standard antibiotics, no conclusions could be made since no statistical testing was conducted and findings of similar subgroups were not reported.83 For example, patients with initial or recurrent CDI participated in the study but the recurrence rate was not reported by the type of CDI as enrollment for the probiotic group.83
Additional Nonstandard Approaches
In addition to the nonstandard interventions for CDI addressed in this review, case reports, or nonexperimental studies reveal numerous other approaches for treating or preventing CDI (See Table 21; Appendix Table C11 is the evidence table). Use of other probiotics (for example, yogurt containing live bacterial cultures)93,189,190 and other cytotoxin absorbing resins191,192 have been reported.
Another approach under investigation for treatment of recurrent or refractory CDI is fecal flora reconstitution, which instills feces from a healthy donor into the colon of a patient with CDI. Six case studies/series have been published,96,97,128-130,193 four within the last 2 years. 128-130,193 Of a total of 60 patients; 52 patients (87 percent) resolved diarrhea and experienced no further relapse during followup. Two studies reported relapse of diarrhea in 7 of 34 patients (21 percent). Followup periods ranged from 3 weeks to 8 years.
Other nonstandard interventions include a monoclonal antibody to C. difficile toxin A,194 intravenous immunoglobulin,106,195-197 two nonstandard antibiotics, Tigecycline,198 a C. difficile toxoid vaccine,199 and a nontoxigenic strain of C. difficile.200
Potential Harms
Harmful effects of nonstandard interventions for CDI appear to be few, but not all studies or case reports included adverse effects in their finding (Tables 19–21). A serious potential harm associated with administration of probiotics for CDI in critically ill patients is fungemia.93,94 In one review of an outbreak, previous medical charts, and the literature, 46 percent of 60 critically ill patients who developed fungemia had been administered a probiotic, and 28 percent subsequently died.93 In addition, McFarland reported finding 12 cases of Lactobacillus bacteremia in patients (mostly children) taking a probiotic containing Lactobacillus.11 Minor adverse symptoms of probiotics and prebiotics were abdominal symptoms such as nausea, bloating, and vomiting, and they have not differed significantly from those of subjects receiving placebo or an active control.11,83,91 Headache (one subject), and abdominal pain, change in bowel habit, and polymyalgia rheumatica (one subject) occurred following C. difficile vaccination.199 Hypotension, diarrhea, headache, nausea, and abdominal discomfort were reported after administration of a monoclonal antibody to C. difficile toxin A.194
- How do Different Methods for Detection of Toxigenic C. difficile Compare in Their Sensitivity and Specificity?
- What are Effective Prevention Strategies?
- What are the Comparative Effectiveness and Harms of Different Antibiotic Treatments?
- What are the Effectiveness and Harms of Nonstandard Adjunctive Interventions?
- Results - Effectiveness of Early Diagnosis, Prevention, and Treatment of Clostri...Results - Effectiveness of Early Diagnosis, Prevention, and Treatment of Clostridium difficile Infection
Your browsing activity is empty.
Activity recording is turned off.
See more...