Randomised Clinical Trial

A 12-Strain Bacterial Mixture Versus Faecal Microbiota Transplantation Versus Vancomycin for Recurrent Clostridioides difficile Infections

Anne Abildtrup Rode; Mahtab Chehri; Laura Rindom Krogsgaard; Kristine Klysner Heno; Anna Tølbøll Svendsen; Iben Ribberholt; Morten Helms; Jørgen Engberg; Kristian Schønning; Michael Tvede; Christian Østergaard Andersen; Ulrich Stab Jensen; Andreas Munk Petersen; Peter Bytzer


Aliment Pharmacol Ther. 2021;53(9):999-1009. 

In This Article

Materials and Methods

Study Design

We compared FMT, rectal bacteriotherapy and standard-of-care oral vancomycin for treatment of recurrent C difficile infection in an open-label, multicentre randomised controlled trial. We planned an interim analysis of the primary endpoint for the first 90 participants.

Study Participants

The trial was conducted at two centres in Denmark. We assessed the eligibility of all consecutive individuals with a positive test for C difficile from all clinical microbiological laboratories in eastern Denmark, including individuals from primary, secondary and tertiary care, covering a population of approximately 2.7 million.

Eligible persons were adults with recurrence of C difficile infection, defined as diarrhoea and a new positive test for C difficile (or pseudomembranous colitis diagnosed by endoscopy or in biopsy) within 90 days after a former episode of C difficile infection. We defined diarrhoea as the passage of ≥3 loose or liquid stools per day.[22] The C difficile test used was a polymerase chain reaction (PCR) test for toxin genes (laboratory developed qPCR or GeneXpert®; Cepheid),[23,24] including genes for C difficile toxin B, the binary toxin gene and the Δ117-deletion in the tcdC-gene characteristic for PCR ribotype 027 (CD027). Further inclusion and exclusion criteria are listed in Table 1.

We allowed participants to have started oral vancomycin within seven days prior to inclusion to avoid excluding cases needing urgent treatment. Intake prior to inclusion was included in the total treatment duration.

If necessary, investigators visited eligible persons at home or at the department of admission to minimise selection bias.


We stratified participants according to number of recurrences (first recurrence vs ≥2 recurrences) and randomly allocated them 1:1:1 to either FMT, rectal bacteriotherapy or oral vancomycin. Computer-generated stratified randomisation in blocks of six was used with allocation concealment in sealed opaque envelopes with sequential numbers for each stratum. An independent party conducted the block size, randomisation code and packing of envelopes. The trial personnel were blinded to this process. After allocation, the trial was open-label.

Faecal Microbiota Transplantation

The treatment regimens for the groups receiving FMT or rectal bacteriotherapy are described and compared in Table 2. After a pre-treatment with oral capsule vancomycin, we administered the FMT by enema once, but with a possibility to repeat it up to two times within 14 days after the first infusion, since repetition improved the success rate in former studies.[7,8] The indication for repetition was ongoing or new-onset diarrhoea (≥3 loose or liquid stools per day), as judged by a trial physician, without new testing for C difficile. We used a different donor when repeating FMTs.

We used frozen donor stool from a donor stool bank with extensively tested (Table S1) universal donors recruited from the Danish Blood Donor Corps.[25] The donor stool bank was described elsewhere.[22]

Rectal Bacteriotherapy

The bacterial mixture for rectal bacteriotherapy consists of 12 well-characterised gut bacterial strains (Table 2). The bacteria were originally isolated from faeces from healthy volunteers. Details on the selection process were described elsewhere.[19,20] All strains are sensitive to either metronidazole or ampicillin. These antibiotics can be used, should a treatment-induced infection be suspected.[19,20]

The bacterial mixture was produced each day of treatment and transported directly to the study site. It is considered a drug, produced following Good Manufacturing Practice principles, and approved and monitored by the Danish Medicines Agency (Drug id: 27415700915).

Participants receiving rectal bacteriotherapy were also pre-treated with oral vancomycin, and the bacterial mixture was administered using the same procedure as for FMT. In contrast to FMT, all participants received three infusion of rectal bacteriotherapy on 3 consecutive days. The 3-day regimen was based on former clinical experience.[20]

Oral Vancomycin

All participants in the vancomycin group received monotherapy with oral capsule vancomycin 125 mg four times daily for 14 days. Furthermore, participants with ≥2 recurrences of CDI were treated with additionally 5 weeks of tapering as recommended in available guidelines when commencing the study.[11] The tapering regimen included oral vancomycin 125 mg twice daily for 1 week, 125 mg once daily for 1 week, 125 mg every other day for 1 week and 125 mg every third day for 2 weeks.


We followed participants by telephone calls and chart reviews on day 1, 3, 7, 14, 30, 90 and 180 after ended treatment. In case of repeating FMT, we restarted follow-up from day 1. If participants reported diarrhoea (as defined above) in the follow-up period, a new C difficile test was performed.

We defined a new episode of C difficile infection as either a clinical treatment failure or a recurrence of C difficile infection. Clinical treatment failure was defined as ongoing, worsened or new-onset diarrhoea during treatment, that is from inclusion to last intake of vancomycin in the vancomycin group; to last infusion in the rectal bacteriotherapy group, and to 14 days after the first infusion in the FMT group, because of the inherent possibility to repeat the treatment. Evaluation of a clinical treatment failure was solely clinical with no confirmatory microbiological tests before initiating or changing treatment. A recurrence was defined as a new episode of C difficile infection, that is new-onset diarrhoea and a positive C difficile test, after ended treatment.

We registered any adverse events and serious adverse events in the period from inclusion to 14 days after ended treatment, using EMAs definitions.[26]


The primary endpoint was clinical cure, defined as absence of C difficile infection (ie absence of diarrhoea or diarrhoea with a negative C difficile test), within 90 days after ended treatment. Pre-specified subgroup analyses included analysis of the primary endpoint according to stratification—first recurrence vs 2 recurrences—and according to toxin profile—that is C difficile strains with the toxin B-gene, C difficile strains containing both the toxin B-gene and the binary toxin gene and C difficile strains of presumptive PCR ribotype 027 (CD027).

Secondary endpoints included clinical cure within 180 days after ended treatment, safety (occurrence of adverse events and serious adverse events) and 180-day mortality (all-cause and possibly C difficile-related mortality).

Statistical Analyses

In a sample size calculation, 73 participants were needed in each group to detect a significant difference (P < 0.05) in cure rates between vancomycin (estimated cure rate of 60%) and FMT/rectal bacteriotherapy (estimated cure rate of 80%) with a power of 80%. A non-inferiority analysis between FMT and rectal bacteriotherapy with a maximal difference in cure rate of 5% required 932 participants in each group with a power of 80%. With these conservative estimates and for feasibility reasons, we planned to include 150 participants in each group, 450 participants in total.

We planned a pre-specified interim analysis on the primary endpoint for the first 90 participants, using the Haybittle-Peto boundary (P < 0.001) to determine if a potentially inferior treatment (if only one) should be removed from the trial or if the trial should be terminated (if two inferior treatments).

We described continuous data with the median (interquartile range and range), compared the three groups using Kruskal-Wallis test, and continued with pairwise Wilcoxon rank sum test if a significant difference was detected. Categorical data were reported as numbers and proportions and compared with odds ratios (OR) (95% confidence intervals), chi-squared test or Fisher's exact test. Furthermore, the primary endpoint was compared in a logistic regression model adjusting for the stratification as appropriate for studies with stratified randomisation.[27,28]

The primary endpoint was analysed on both intention-to-treat, modified intention-to-treat and per-protocol basis.

We used a statistical significance level of 0.05 for all analyses. We performed all statistical analyses in R studio version 1.1.463. Figures were made using GraphPad Prism version 8.2.


All participants gave written informed consent. The Regional Committee of Health Research Ethics (SJ-478), the Danish Medicines Agency (2015-003062-82) and The Danish Data Protection Agency (REG-103-2015) approved the study. We conducted the trial according to the Helsinki Declaration II and to the principles of Good Clinical Practice. An external party monitored the trial. The study was registered at EudraCT (2015-003062-82) and ClinicalTrials.gov (NCT02774382).