Summary of the evidence

This is a summary of the evidence, for full details (including the economic analysis) see the evidence review.

The evidence for treating Clostridioides difficile infection in adults specifically included antibiotic efficacy, choice, dose and dose frequency, faecal microbiota transplantation (FMT), bezlotoxumab and prebiotics. The evidence for treating C. difficile infection in children included antibiotic choice and probiotics.

For C. difficile infection in adults, young people or children, no evidence from systematic reviews or randomised controlled trials (RCTs) was identified for antibiotic prescribing strategies, course length or route of administration. There was also no evidence found for probiotics for C. difficile infection in adults, nor for antibiotic efficacy, dose or dose frequency, FMT, bezlotoxumab or prebiotics for infection in children.

There was evidence found for prophylactic antibiotics (in adults having a stem cell transplant or in hospital), prebiotics and probiotics to prevent C. difficile infection in adults. There was evidence for probiotics to prevent C. difficile infection in children.

Interventions included in the search were antimicrobial interventions, non-antimicrobial interventions (bezlotoxumab and intravenous immunoglobulin), and non-pharmacological interventions (probiotics, prebiotics, FMT, and stopping current antibiotics or proton pump inhibitors). No evidence from systematic reviews or RCTs was found for intravenous immunoglobulin or stopping current antibiotics or proton pump inhibitors. In addition, the following interventions were outside the scope of this guideline because there is no UK licensed product available: ridinilazole, cadazolid, surotomycin, nitazoxanide, tolevamer, LFF517, bacitracin and tolevamer.

Treating initial or first recurrent C. difficile infection in adults

Antibiotics

Antibiotic efficacy

A statistically significant improvement was seen in symptomatic and bacteriological cure with vancomycin 125 mg four times daily for 5 days compared with placebo in adults with first-episode pseudomembranous colitis (some associated with evidence of C. difficile infection; Nelson et al. 2017).

Antibiotic choice

In 1 network meta-analysis, different antibiotic treatments were compared for treating the initial or first recurrent episode of C. difficile infection. Vancomycin was used as the reference treatment (Beinortas et al. 2018), and the treatments were ranked using P scores. Of the antibiotics available in the UK, sustained symptomatic cure was most effective with teicoplanin (P score=0.9386), followed by fidaxomicin (P score=0.7922), vancomycin (P score=0.4850), rifaximin (P score=0.4296), fusidic acid (P score=0.3794) and metronidazole (P score=0.2411). P scores are calculated as the average p value for superiority for that intervention compared with all the other interventions in the network meta-analysis. They take account of the magnitude of the difference and the level of uncertainty. Higher P scores (on a 0 to 1 scale) represent treatments in which there is more confidence that they are better than the other alternatives in the network meta-analysis.

A sensitivity analysis was done in which the effect was explored of removing studies with fewer than 50 people per arm, studies that were published before 2000, and unblinded studies. When non-blinded studies or studies with fewer than 50 people per arm were removed, fidaxomicin was the highest ranked treatment available in the UK. When studies published before the year 2000 were removed, teicoplanin was the highest ranked treatment available in the UK, followed by fidaxomicin.

Subgroup analysis was done for severe C. difficile infection, non-severe C. difficile infection, initial C. difficile infection, non-initial C. difficile infection, people aged 65 years and over and people aged under 65 years. For all subgroups, fidaxomicin was the highest ranked treatment available in the UK, and metronidazole was the least effective (being ranked either the fifth, sixth or seventh most effective option in the different subgroups).

There were no statistically significant differences in clinical effectiveness (recurrence of C. difficile infection, clinical resolution of C. difficile infection, relapse of C. difficile infection at 5 weeks and adverse events) for oral vancomycin compared with fidaxomicin (Hvas et al. 2019).

Antibiotic dose

There was no statistically significant difference in clinical effectiveness (symptomatic cure) with low-dose (125 mg four times a day) compared with high-dose (500 mg four times a day) vancomycin, both taken for 5 to 15 days (Nelson et al. 2017).

There was a statistically significant improvement in clinical effectiveness (symptomatic cure) with fidaxomicin 400 mg daily compared with fidaxomicin 100 mg or 200 mg daily, all taken for 10 days.

Antibiotic dose frequency

There was no statistically significant difference in clinical effectiveness (symptomatic cure) with 100 mg of teicoplanin twice daily compared with 50 mg of teicoplanin four times daily (Nelson et al. 2017).

FMT for treating initial C. difficile infection

There were no statistically significant differences in clinical effectiveness (resolution of C. difficile infection, treatment failure, all-cause and C. difficile infection attributable mortality or length of stay) of:

  • the first dose of FMT compared with vancomycin

  • the second dose of FMT compared with vancomycin (Camacho-Ortiz et al. 2017).

FMT for treating recurrent C. difficile infection

There were statistically significant increases in clinical effectiveness (resolution of symptoms, resolution of diarrhoea, relapse of diarrhoea) with:

  • a 4- to 10‑day course of vancomycin followed by FMT compared with 10 days of vancomycin at 1- and 8‑week follow up (Hvas et al. 2019)

  • a 4- to 10‑day course of vancomycin followed by FMT compared with 10 days of fidaxomicin at 8‑week follow up (Hvas et al. 2019)

  • a 4- to 5‑day course of vancomycin plus bowel lavage followed by FMT compared with either 14 days of vancomycin (with or without bowel lavage) at 10‑week follow up, and at 5‑week follow up for relapse (van Nood et al. 2013)

  • a 3‑day course of vancomycin followed by FMT compared with a standard then a pulsed course of vancomycin at 10‑week follow up (Cammarota et al. 2015).

There were no statistically significant differences in all-cause or C. difficile infection-related mortality for a short course of vancomycin plus bowel lavage followed by FMT compared with 14 days of vancomycin or 14 days of vancomycin plus bowel lavage (van Nood et al. 2013).

There were no statistically significant differences in adverse events for:

  • a short course of oral vancomycin followed by FMT compared with either 10 days of vancomycin or fidaxomicin (Hvas et al. 2019)

  • a short course of vancomycin followed by FMT compared with vancomycin, either with or without bowel lavage (van Nood et al. 2013).

There was a statistically significant lower mean number of days of diarrhoea compared with a course of vancomycin followed by FMT compared with tapered vancomycin (Hota et al. 2017). However, a short course of vancomycin followed by FMT or bowel lavage plus FMT statistically significantly increased treatment-related diarrhoea, bloating or cramping (Cammarota et al. 2015; van Nood et al. 2013).

Serious adverse events were reported in 2 RCTs. In 1 RCT, a sepsis-like response occurred (possibly related to FMT) but resolved without admission or treatment (Hvas et al. 2019). In the other RCT, 3 serious adverse events were noted but none were thought to be treatment related (Hota et al. 2017).

Preventing recurrence in people with C. difficile infection in adults

Antibiotics

In adults who had an initial or first recurrent episode of C. difficile infection treated with vancomycin or metronidazole, immediate rifaximin for 20 days was statistically significantly more effective than placebo at reducing recurrence of both C. difficile infection-confirmed diarrhoea and self-reported diarrhoea. However, when the outcomes of recurrent C. difficile infection-confirmed diarrhoea and recurrent self-reported diarrhoea were analysed separately, there was no statistically significant difference between rifaximin and placebo in either group (Garey et al. 2011).

In adults who had an initial, first recurrent, or second or later recurrent episode of C. difficile infection treated with vancomycin or metronidazole, there was no statistically significant difference between immediate rifaximin for 28 days and placebo for recurrent C. difficile infection at 12 weeks or 6 months, or for rehospitalisation for C. difficile infection within 6 months. When subgroup analysis was done for standard care antibiotic treatment with metronidazole or vancomycin, there was no statistically significant difference between rifaximin and placebo for C. difficile infection recurrence. There was also no statistically significant difference in effect between rifaximin and placebo on C. difficile infection recurrence when post‑hoc analyses were done for C. difficile infection history (Major et al. 2019).

A Kaplan–Meier analysis showed that rifaximin led to a statistically significant increased time to both recurrent C. difficile infection-confirmed diarrhoea and recurrent self-reported diarrhoea compared with placebo (Garey et al. 2011). However, when the time to C. difficile infection-confirmed diarrhoea and time to self-reported diarrhoea were analysed separately, there was no statistically significant difference between rifaximin and placebo.

There were no statistically significant differences between rifaximin and placebo for mortality, serious and non-serious adverse events (Major et al. 2019).

Monoclonal antibodies

In adults with an initial or recurrent episode of C. difficile infection treated with standard care antibiotic treatment (that is, metronidazole, vancomycin or fidaxomicin), bezlotoxumab was statistically significantly more effective than placebo for recurrent C. difficile infection, 12 weeks of sustained cure and recurrence of diarrhoea (regardless of whether it was associated with a positive toxin test; Wilcox et al. 2017). A Kaplan–Meier analysis suggested that bezlotoxumab increased time to recurrence of C. difficile infection compared with placebo, but it was unclear if the differences were statistically significant.

Various subgroup analyses for C. difficile infection risk factors and stratification variables were done. Bezlotoxumab was statistically significantly more effective than placebo for recurrence of C. difficile infection for the stratification variables of inpatients and outpatients, and whether people had vancomycin or metronidazole as their standard care antibiotic treatment. However, there was no statistically significant difference between bezlotoxumab and placebo for the outcome of recurrence of C. difficile infection for the stratification variable of people having fidaxomicin as their standard care antibiotic treatment.

There were no statistically significant differences between bezlotoxumab and placebo for the outcomes of initial clinical cure at 2 days and mortality.

There was no statistically significant difference between bezlotoxumab and placebo for infusion-specific adverse events or adverse events leading to treatment being stopped at 24‑hour follow up. There was also no statistically significant difference between bezlotoxumab and placebo for drug-related adverse events, other adverse events (most commonly abdominal pain, diarrhoea, nausea, vomiting, fatigue, pyrexia, serious C. difficile, urinary tract infection or headache), serious adverse events or for drug-related serious adverse events, occurring during the 4 weeks after the bezlotoxumab infusion.

FMT for preventing C. difficile infection recurrence

In NICE analyses, there were no statistically significant differences in the clinical effectiveness (recurrence) of the following doses of FMT taken after antibiotic treatment for a current episode of C. difficile infection in adults with multiple recurrent infections:

  • a single dose of FMT compared with placebo

  • 2 doses of FMT compared with placebo

  • 2 doses of FMT compared with a single dose of FMT

  • 1 or 2 doses of FMT (pooled) compared with placebo (Dubberke et al. 2018).

There was no statistically significant difference in adverse events. However, 3 severe adverse events were reported and thought to be related to FMT in the '2 doses of FMT' group. There were 6 deaths (3 in the '2 doses of FMT' group and 3 in the '1 dose of FMT' group) in the FMT arms of the trial and none in the placebo group.

Prebiotics for relapse of diarrhoea

There was a statistically significant decrease in relapse of diarrhoea with metronidazole or vancomycin plus the prebiotic oligofructose compared with metronidazole or vancomycin plus placebo for diarrhoea associated with C. difficile infection in adults aged over 65 years (Lewis et al. 2005a). No statistically significant difference was noted for C. difficile culture positivity at 30- or 60‑day follow up.

Treating initial or recurrent C. difficile infection in children and young people

Antibiotics choice

Oral metronidazole compared with oral rifaximin

There was no statistically significant difference in clinical effectiveness (C. difficile infection cure rate or recurrent C. difficile infection) with oral metronidazole compared with oral rifaximin for a first episode of C. difficile infection in children with inflammatory bowel disease (Gawronska et al. 2017).

Oral fidaxomicin compared with oral vancomycin

There was no statistically significant difference in confirmed clinical response or resolution of diarrhoea with oral fidaxomicin compared with oral vancomycin for confirmed C. difficile infection in children and young people aged under 18 years (Wolf et al. 2020).

In the total study population and subgroup of children aged under 2 years, there was no statistically significant difference between oral fidaxomicin and oral vancomycin for the outcome of global cure. However, in other subgroups (those aged 2 years and over and those with a positive toxin test aged 2 years and over), fidaxomicin was statically significantly more effective than vancomycin for global cure.

Oral fidaxomicin statistically significantly reduced C. difficile infection recurrence compared with oral vancomycin in children and young people aged under 18 years. When results were stratified by age, fidaxomicin was statistically significantly more effective than vancomycin in children aged 2 years and over and in children with a positive toxin test aged 2 years and over, but the effect was no longer statistically significant in those aged under 2 years.

There was no statistically significant difference for treatment-emergent adverse events (including serious events, drug-related events, those leading to death or withdrawal from treatment).

Probiotics for persistent diarrhoea

There was a statistically significant reduction in the mean number of days of diarrhoea with oral rehydration solution plus the probiotic Lactobacillus rhamnosus GG compared with oral rehydration solution alone in children with a positive C. difficile stool culture (Basu et al. 2007). However, there was no statistically significant difference in the mean number of days of vomiting.

Preventing C. difficile infection in adults without infection

Antibiotics

In people without C. difficile infection having a haematopoietic stem cell transplant and fluoroquinolone prophylaxis during neutropenia, there was no statistically significant difference between fidaxomicin and placebo for reducing prophylaxis failure at 30, 60 or 70 days (Mullane et al. 2019). There was also no statistically significant difference between fidaxomicin and placebo for any adverse events reported in the study.

Fidaxomicin was statistically significantly more effective than placebo at reducing confirmed diarrhoea associated with C. difficile infection at 30, 60 and 70 days. A Kaplan–Meier analysis showed a statistically significantly increased time to recurrence of C. difficile infection with fidaxomicin compared with placebo.

In people without C. difficile infection who were hospitalised for up to 30 days before their current hospitalisation, there was no statistically significant difference between oral vancomycin and placebo for:

  • healthcare facility-onset (symptomatic infection more than 72 hours after hospital admission) C. difficile infection, or

  • community-onset healthcare facility-associated (symptomatic infection up to 3 months after hospital discharge) C. difficile infection after hospital discharge (Johnson et al. 2020).

Prebiotics

In inpatients aged over 65 years without C. difficile infection who were prescribed a broad-spectrum antibiotic, the prebiotic oligofructose did not have a statistically significantly different effect to placebo at end of follow up for all-cause mortality or for incidence of diarrhoea, significant diarrhoea (3 loose stools or more in a 24‑hour period), non-significant diarrhoea (1 or 2 loose stools in a 24‑hour period), C. difficile-associated diarrhoea or C. difficile-associated significant diarrhoea (Lewis et al. 2005b).

In the oligofructose group, the median (interquartile range) length of hospital stay was 17 days (13 to 22) compared with 15 days (11 to 18) in the placebo group.

Probiotics

The evidence for probiotics in the prevention of C. difficile infection in adults comes from 1 systematic review (Goldenberg et al. 2017). The population in the included studies was people aged over 18 years having antibiotic treatment for any reason.

Probiotics statistically significantly reduced the incidence of C. difficile infection compared with any comparator (follow-up time point not reported) in studies in inpatients, but not in studies in outpatients or patients in mixed settings.

Probiotics were not statistically significantly different compared with any comparator for the outcome of incidence of C. difficile infection determined by detection of C. difficile in stools, either overall or in any setting (inpatients, outpatients, or mixed settings; follow-up time points not reported).

Probiotics statistically significantly reduced the number of adverse events compared with any comparator (follow-up time point not reported). Details of the adverse events were not reported.

Preventing C. difficile infection in children and young people without infection

Probiotics

The evidence for probiotics in preventing C. difficile infection in children and young people comes from 1 systematic review (Goldenberg et al. 2017) and 1 RCT (Kolodziej and Szajewska 2019). The population in the included studies was children and young people aged under 18 years who were having antibiotic treatment for any reason.

Probiotics statistically significantly reduced the incidence of C. difficile infection compared with any comparator (follow-up time point not reported) in the inpatient and mixed settings studies.

Probiotics were not statistically significantly different compared with any comparator in inpatient studies for the outcome of incidence of C. difficile infection determined by detection of C. difficile in stool (follow-up time point not reported).

Probiotics were not statistically significantly different compared with any comparator for adverse events.

  • National Institute for Health and Care Excellence (NICE)