PubMed Health. A service of the National Library of Medicine, National Institutes of Health.

Perras C, Tsakonas E, Ndegwa S, et al. Vancomycin or Metronidazole for Treatment of Clostridium difficile Infection: Clinical and Economic Analyses [Internet]. Ottawa (ON): Canadian Agency for Drugs and Technologies in Health; 2011 Jan. (CADTH Technology Report, No. 136.)


4.1. Methods

4.1.1. Literature search

The clinical literature search was performed by an information specialist using a peer-reviewed search strategy.

The following bibliographic databases were searched through the Ovid interface: MEDLINE, MEDLINE In-Process & Other Non-Indexed Citations, EMBASE, Biosis Previews, The Cochrane Library, and the Centre for Reviews and Dissemination databases. The search strategy comprised controlled vocabulary, such as the National Library of Medicine’s MeSH (Medical Subject Headings), and keywords. The main search concepts were vancomycin, metronidazole, and C. difficile. The clinical search was not restricted by publication date, but was restricted to English and French language publications. Methodological filters were applied to limit retrieval to systematic reviews, randomized controlled trials, controlled clinical trials, and observational studies. See Appendix 2 for the detailed search strategies.

The search was run on October 28, 2009. Regular alerts were established to update the search until the publication of the final report.

Grey literature (literature that is not commercially published) was identified by searching the websites of health technology assessment and related agencies, professional associations, clinical trials registries, and other specialized databases. Appendix 2 shows a list of the main grey literature resources. Google and other Internet search engines were used to search for additional information. These searches were supplemented by handsearching the bibliographies and abstracts of key papers and conference proceedings, and through contacts with appropriate experts and agencies.

Three manufacturers (Sanofi-Aventis Canada Inc., Ferring Pharmaceuticals Canada, and Iroko International LP) were contacted to request unpublished clinical studies.

4.1.2. Selection criteria

To be included, the clinical studies had to meet the criteria shown in Table 2.

Table 2. Selection Criteria for Clinical Review.

Table 2

Selection Criteria for Clinical Review.

The original research protocol was followed. After the literature was searched, it was determined that none of the retrieved studies met the population inclusion criteria (none of the studies included only patients with an initial episode of moderate or severe CDI). Some studies included patients with moderate or severe CDI in a mixed population of patients, with an initial or recurrent episode of CDI. One study included only patients with an initial episode of CDI, and mild, moderate, or severe disease. It was decided to proceed with a systematic review in which this information could be used. As a result, the original research questions could not be answered. By broadening the scope, the results would apply to a population that included patients with an initial or a recurrent episode of moderate or severe CDI, instead of only patients with an initial episode.

4.1.3. Selection method

Two reviewers (CP, SN) independently screened the titles and abstracts of all citations that were retrieved in the literature search. Based on the selection criteria that were specified before the research was done, the full text of any articles that met the criteria was ordered. The reviewers then independently reviewed the full text of selected articles, applying the selection criteria, and comparing the included and excluded studies. Disagreements were resolved through discussion until consensus was reached. When consensus could not be reached, a colleague (BH) evaluated the study to determine inclusion. When the full study was retrieved and was deemed not to meet the inclusion criteria, the reason for the exclusion was recorded. Duplicate publications of the same trial were excluded.

4.1.4. Data extraction strategy

A data extraction form was designed before the research was done to document the characteristics and outcomes of the selected studies (Appendix 3). Data were extracted by two independent reviewers (CP, SN), and any disagreement was resolved through discussion until consensus was reached. Where data were insufficient, missing from the research report, or if the study was only available as an abstract or conference proceeding, the corresponding author of the study was contacted to determine if these data could be obtained.

4.1.5. Strategy for validity assessment

A quality assessment of full-text publications of the included studies measuring the clinical effectiveness of vancomycin or metronidazole was independently conducted by two reviewers (CP, SN). The two reviewers resolved any disagreement through discussion until consensus was reached. The methodological quality of the systematic reviews was assessed using the Oxman and Guyatt scale.102 The methodological quality of RCTs and observational studies was assessed using the Downs and Black checklist.103 Guidelines were evaluated using the Appraisal of Guidelines for Research and Evaluation (AGREE) instrument104 (Appendix 4).

4.1.6. Data analysis methods

For RCTs, the relative risks (RR) with corresponding 95% CI were calculated using the Confidence Interval Analysis or CIA software.105 Because of the clinical heterogeneity among studies, a meta-analysis was not performed.

For observational studies, the RRs for prospective studies or the odds ratios (OR) for retrospective studies, with corresponding 95% CI, were calculated using the Confidence Interval Analysis software.105 These results were used in a narrative review.

The findings were interpreted in light of the heterogeneity of the studies (differences in design, study populations, interventions or exposures, and outcome measures) and the quality assessment (confounding, bias, and external validity).

4.2. Results

4.2.1. Quantity of research available

Of the 87 potentially relevant reports that were retrieved for a full text review, 67 were excluded, leaving 20 reports (two systematic reviews, five RCTs, and 13 observational studies) that compared vancomycin and metronidazole in patients with CDI (Figure 1). The excluded clinical studies appear in Appendix 5. None of the 20 reports fully met the original inclusion criteria. Thus, the original research questions could not be answered using the identified literature.

Figure 1. Selected Reports.

Figure 1

Selected Reports. RCTs = randomized controlled trials; SRs = systematic reviews.

No additional studies were provided from the three manufacturers that were contacted.

4.2.2. Study characteristics

a. Systematic reviews

Two systematic reviews66,106 evaluated RCTs comparing antibiotics for the treatment of CDI. The population included patients with CDI and was not limited by the number of episodes or the severity of disease. All the included RCTs that compared metronidazole and vancomycin were conducted before the emergence of the NAP1 strain. The details appear in Appendix 6, Tables 1 and 2.

b. Randomized controlled trials

Of the five RCTs that compared vancomycin to metronidazole,107111 three were available as full text publications,107109 one was available as an abstract,110 and another was available as a conference poster.111 Data were extracted from all five.

All five RCTs included a mixed population of patients with first or recurrent episodes of mild, moderate, or severe CDI. One study provided separate data on patients with moderate disease,111 two RCTs provided data on patients with PMC at diagnosis (a marker for severe disease),107,108 and two RCTs provided data on patients with severe disease.109,111 Patients were hospitalized in three studies.107109 It was unclear whether or not patients were hospitalized in the other two studies.110,111 One RCT111 was conducted during the NAP1 epidemic.

c. Observational studies

Eight retrospective112119 and five prospective120124 studies compared vancomycin to metronidazole. Five of these studies were available as abstracts,116,117,122124 and eight were available as full text publications.112115,118121 Of the thirteen studies, one study123 presented data on patients with an initial episode of CDI after the emergence of the NAP1 strain, and one study112 provided data on patients with moderate or severe disease before the NAP1 epidemic.

4.2.3. Data analyses and synthesis

We extracted data from the five RCTs and the 13 observational studies because this had not been previously done systematically. The study characteristics, results, and quality assessments appear in Tables 1 to 5 of Appendix 7 and Tables 1 to 4 of Appendix 8, respectively. The definitions of disease severity and clinical outcomes that were used in the studies appear in Tables 1 and 2 of Appendix 8.

The effectiveness of vancomycin compared to that of metronidazole in the population that was identified in the research question could not be determined based on the selected studies. The data on patients with moderate CDI, severe CDI, or PMC at diagnosis in a mixed population (those with initial and recurrent episodes), and data on adult patients with an initial episode of CDI of varying disease severity, were analysed.

a. Randomized controlled trials

From the RCTs that provided data on patients with moderate and severe disease, we calculated RR, RR reduction (RRR), or RR increase (RRI), and numbers needed to treat (NNT) (Table 3).

Table 3. Effect Measures in Randomized Controlled Trials.

Table 3

Effect Measures in Randomized Controlled Trials.

Zar et al.109 stratified patients into mild and severe disease groups based on a severity assessment score that was developed for the study. A total of 69 patients were classified as being in the severe disease group. Louie et al.111 presented data on clinical success by CDI severity. In this study, 90 patients were classified as having severe CDI. These two studies showed that, in patients with severe CDI, the use of vancomycin increased the cure rate by 27% (RR 1.27 [95% CI, 1.05 to 1.53]; NNT 5)109 and 31% (RR 1.31 [95% CI, 1.03 to 1.66]; NNT 5)111 compared with metronidazole. These results include patients with initial or recurrent CDI. In Louie et al.’s study,111 more than 70% of patients had an initial episode of CDI.

The results of the severe CDI groups in the Zar and in the Louie studies were not combined in a meta-analysis because the patients in Zar trial were enrolled before the epidemic (1994 to 2002) and the patients in Louie trial were enrolled during the epidemic, with up to a third of the patients testing positive for NAP1 (Thomas Louie, University of Calgary, Calgary, AB: personal communication, 2010 February 24). Furthermore, different criteria were used to define disease severity. In Zar et al.’s trial,109 patients were included in the severe disease group if they had two of the following: age over 60 years, a temperature greater than 38.3°C, an albumin level of less than 2.5 mg/dL, or a peripheral white blood cell count greater than 15,000 cells/mm3 within 48 hours of enrollment; or one of the following: endoscopic evidence of PMC, or treatment in the ICU. In Louie et al.’s trial,111 patients met the following three criteria to be included in the severe group: 10 or more bowel movements per day, a white blood cell count of greater than 20,000/mm3, and severe abdominal pain.

Teasley et al.107 and Wenisch et al.108 provided subgroup analyses of patients with PMC, a marker for disease severity. The relative risks of recurrences in patients with PMC at diagnosis were not calculated because of the small number of events.107,108 Furthermore, Teasley et al.’s and Wenisch et al.’s trials were not combined in a meta-analysis because the two populations were not clinically homogeneous.107,108 The patients in Teasley et al.’s trial were selected from a veteran’s medical centre and were more than 20 years older than the patients in Wenisch et al.’s trial. They had underlying diseases and comorbidities, and more than 80% had undergone surgery during admission compared to less than half of the patients in Wenisch et al.’s trial.

Zar et al.’s RCT109 reported the incidence of colectomy and all-cause mortality. The relative risks were not calculated for these outcomes because of the small number of events.

Other effect measures were calculated (Table 3), but the comparisons between vancomycin and metronidazole produced inconclusive findings.

The Downs and Black criteria103 were used to assess study quality and limitations. In Teasley et al.’s study107 and Zar et al.’s study,109 a lower dose of metronidazole than that recommended in recent guidelines was used (1 g total daily dose instead of 1.5 g).19 In Zar et al.’s study,109 the drop-out rate was 12.8% among all randomized patients and 15.8% among those with severe disease. Among patients with severe CDI, seven patients on vancomycin and six patients on metronidazole did not complete the treatment course and were excluded from the analysis. Of these 13 patients, seven died within five days of therapy. The analysis was not based on intention to treat, and the cure rate may have been overestimated. The drop-out rate was 36.4% among all patients who were randomized in Louie et al.’s trial.111 When excluding the patients on tolevamer, a drop-out rate of 19.8% in the vancomycin and metronidazole groups combined was reported. Separate drop-out data were not provided for patients with severe disease. Teasley et al.’s study107 and Wenisch et al.’s study108 were neither placebo-controlled nor double-blinded, and the concealment of treatment allocation was not described. This may have biased the results (selection and performance biases).

b. Observational studies

The effect measures were calculated for two observational studies,112,123 but the comparisons between vancomycin and metronidazole yielded inconclusive findings (Table 4).

Table 4. Effect Measures Observational Studies .

Table 4

Effect Measures Observational Studies .

Talbot et al.’s112 study is the only one that included children (16 participants) in the study population. No separate information (severity of disease, treatments, or results) was provided for the children.

It could not be determined whether or not the identified RCTs and observational studies included patients who were treated with vancomycin or metronidazole as outpatients in the community.

Copyright © 2011 Canadian Agency for Drugs and Technologies in Health.

Except where otherwise noted, this work is distributed under the terms of a Creative Commons Attribution-NonCommercial- NoDerivatives 4.0 International licence (CC BY-NC-ND), a copy of which is available at

Cover of Vancomycin or Metronidazole for Treatment of Clostridium difficile Infection: Clinical and Economic Analyses
Vancomycin or Metronidazole for Treatment of Clostridium difficile Infection: Clinical and Economic Analyses [Internet].
CADTH Technology Report, No. 136.
Perras C, Tsakonas E, Ndegwa S, et al.


In this Page

PubMed Health Blog...

read all...

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...