3 Evidence

NICE commissioned an external assessment centre (EAC) to review the evidence. This section summarises that review. Full details of all the evidence are in the project documents on the NICE website.

Clinical evidence

The clinical evidence includes 5 randomised controlled trials

3.1 The EAC did a systematic review to identify randomised controlled trials (RCTs) comparing faecal microbiota transplant (FMT), by any route of delivery, with NICE-recommended comparators, to treat a Clostridioides difficile infection in people who have had at least 2 previous episodes. It identified and assessed 5 eligible RCTs including 274 adults in total. These trials compared FMT, given via different routes of administration and with a preceding course of antibiotics, with antibiotic treatment. For full details of the clinical evidence, see section 3 of the assessment report.

More C. difficile infections were resolved with FMT than antibiotic treatment in 4 RCTs; there was no difference in 1 RCT

3.2 FMT (with pretreatment antibiotics) was significantly better at resolving a C. difficile infection than:

  • vancomycin in 4 RCTs (Cammarota et al. 2015, Hvas et al. 2019, Rode et al. 2021 and van Nood et al. 2013)

  • fidaxomicin in 1 RCT (Hvas et al. 2019).

    C. difficile
    infection was resolved in 57% (Rode et al. 2021) to 94% (van Nood et al. 2013) of people having FMT (when any number of infusions was considered). In comparison, C. difficile infection was resolved in 19% (Hvas et al. 2019; vancomycin) to 46% (Rode et al. 2021; vancomycin taper pulse [VTP]) of people having antibiotic therapy in these RCTs. However, Hota et al. (2017) showed less C. difficile infection resolution in the FMT group (given via enema; 43.8%) compared with VTP (58.3%), although the study did not report statistical significance.

Recurrence rate is comparable to or lower than antibiotic treatment

3.3 Three trials found lower C. difficile infection recurrence in the FMT group (range 6% to 10%) compared with the antibiotic group (vancomycin range 62% to 69%, fidaxomicin 46%; Cammarota et al. 2015, Hvas et al. 2019 and van Nood et al. 2013). Hota et al. (2017) reported comparable C. difficile infection recurrence after FMT by enema (56.2%) and VTP (41.7%). However, none of the trials reported statistical significance.

Gastrointestinal side effects can occur in the short term after FMT treatment

3.4 Short-term gastrointestinal side effects were reported in 4 RCTs (Cammarota et al. 2015, Hota et al. 2017, Hvas et al. 2019 and van Nood et al. 2013). The most common effects included diarrhoea, bloating, abdominal pain or cramps. These symptoms lasted (when reported) between 3 hours (van Nood et al. 2013) and 12 hours (Cammarota et al. 2015), or were described as 'transient' (Hvas et al. 2019).

Small sample sizes and the relevance of the population to the NHS limit the evidence

3.5 The included studies had relatively small sample sizes, with a median of 39 and a range of 27 (Rode et al. 2021) to 64 adults (Hvas et al. 2019). This was partly because 4 of the trials stopped early; only 1 completed after recruiting the target number of people (Hvas et al. 2019). The evidence is also limited by not being done in the UK and the trial populations having fewer comorbidities and a lower chance of being hospitalised than the eligible UK population.

Heterogeneous study characteristics may limit the evidence

3.6 The included studies used different FMT administration routes:

  • 2 used an enema (Hota et al. 2017 and Rode et al. 2021)

  • 1 used colonoscopy (Cammarota et al. 2015)

  • 1 used a nasoduodenal tube (NDT; van Nood et al. 2013)

  • 1 used mixed routes (colonoscopy or nasojejunal tube; Hvas et al. 2019).

    None of the included trials evaluated FMT delivered via capsule, nasogastric tube (NGT) or flexible sigmoidoscopy. The number of times FMT was given also varied, from 1 to 4 infusions. In 3 of the trials a proportion of people taking part were being treated for a first recurrence of C. difficile infection (Cammarota et al. 2015, Hota et al. 2017 and van Nood et al. 2013). However, this was only a minority of cases.

Cost evidence

Of 8 economic studies found, 1 used an NHS perspective

3.7 The EAC did a systematic review to find economic evaluations comparing FMT, by any route of delivery, with NICE-recommended comparators, to treat a C. difficile infection in people who have had at least 2 previous episodes. It assessed 8 economic evaluation studies relevant to the decision problem. Abdali et al. (2020) was a UK-based cost–utility analysis comparing 4 treatments for recurrent C. difficile infection (FMT via NGT, FMT via colonoscopy, oral fidaxomicin, and oral vancomycin). The analysis used a Markov model with 4 health states (relapsed, recovered, recurrent C. difficile infection and dead) and had a cycle length of 2 months and time horizon of 1 year. The analysis found that fidaxomicin and vancomycin were dominated by FMT via NGT and FMT via colonoscopy (that is, FMT was cost saving and more effective).

For full details of the cost evidence, see section 4 of the assessment report and the assessment report appendix.

A Markov model compared FMT with antibiotic treatment

3.8 The EAC created a cohort Markov model that included adults with recurrent C. difficile infection who have had 2 or more previous episodes. It had a time horizon of 6 months and cycle length of 2 months. The model included 4 routes of FMT administration (colonoscopy, enema, NDT and oral capsules) and 3 antibiotic comparators (vancomycin, fidaxomicin and VTP). It had 4 health states:

  • recurrent C. difficile infection (starting state)

  • persistent C. difficile infection (recurrent, relapsed or refractory C. difficile infection)

  • recovered

  • dead.

The quality of the clinical evidence limits the economic model

3.9 The quality of the clinical evidence leads to uncertainty in the clinical parameters used in the economic model. No eligible RCTs were identified comparing FMT oral capsules with antibiotics in people with 2 or more previous episodes of C. difficile infection. However, because 2 studies found oral capsules were comparable to FMT colonoscopy (Kao et al. 2017, Ramai et al. 2020) the EAC assumed the transition probabilities to be the same. FMT via NGT and flexible sigmoidoscopy were excluded from the economic model because of a lack of RCT-level data from the clinical evidence review.

The economic model used a number of clinical assumptions

3.10 The economic model used the following clinical assumptions:

  • people are treated with the same treatment again if the first treatment does not work

  • there are constant treatment response and recurrence rates in each cycle

  • pretreatment with antibiotics is only used for the first FMT treatment

  • for anyone in the recovered group there is the same risk of death as the general population

  • initial treatment includes 5 days of hospital stay for FMT and 10 days for antibiotics

  • costs of tests and follow up are assumed to be the same between groups and so are excluded from the model.

    After talking to clinical experts, the assumptions around hospital stay and pretreatment antibiotics were amended in the base case, as discussed in section 4.9 and section 4.10.

FMT by all administration routes evaluated was cost saving in the base case

3.11 The EAC's base case analysis found that all 4 routes of FMT were associated with increased health benefits and reduced costs against all 3 antibiotic comparators, with savings ranging from £3,369 (FMT enema compared with VTP) to £13,134 (FMT oral capsule compared with vancomycin). Health benefits ranged from a quality-adjusted life year (QALY) gain of 0.17 (FMT enema compared with VTP) to 0.66 (FMT via NDT compared with vancomycin).

FMT via NGT could also be cost saving, although there is no RCT-level evidence

3.12 The EAC identified a meta-analysis by Ramai et al. (2020), which suggested an overall cure rate of 78.1% when FMT is given via NGT. The cost of delivering FMT via NGT is estimated to be £740 (Abdali et al. 2020). Because the cure rate is estimated to be higher for FMT via NGT than via enema, and costs less, FMT via NGT is likely to be cost saving for recurrent C. difficile infections, against all 3 comparators considered.

FMT remained cost saving in the sensitivity and scenario analyses

3.13 The EAC did deterministic and probabilistic sensitivity analyses, and scenario analyses. The deterministic sensitivity analysis compared FMT via enema (the least cost saving FMT route) with VTP (the comparator with the second lowest cost and highest health benefit). It found that the largest cost drivers were the resolution probability for FMT via enema and VTP, followed by the hospital stay for any cases of C. difficile infection in subsequent cycles. The results of the probabilistic sensitivity analysis showed that FMT is estimated to be cost saving 96% to 100% of the time compared with antibiotic treatment. The EAC also did 5 scenario analyses. FMT remained cost saving in all of them:

  • Pretreatment antibiotics for all FMT treatments, not just the index treatment (FMT was compared with the VTP treatment group only).

  • Subsequent treatment with VTP for all treatment arms for people in the persistent C. difficile infection state.

  • Threshold analysis on fidaxomicin cost discounting.

  • Extending the time horizon from 6 months to 1 year.

  • All treatment arms having a 1-day hospital stay for the index treatment instead of 5 or 10 days' stay in the FMT and antibiotic groups, respectively.

    For full details see assessment report appendix 1 in the supporting documents for this guidance.