Comment: A recent systematic review found that FMT was used to treat an initial episode of CDI in a very small (n=7) number of patients, with discrepant results. 6 These data come from two small case series of patients with refractory or severe first episode of CDI. 33 , 34 In addition, a preliminary report showed that treatment with FMT after the first episode of CDI can decrease mortality in patients infected with the 027 strain in comparison to FMT performed after recurrence. 29 However, additional studies will need to establish if FMT is more cost-effective than antibiotics for the treatment of the first episode of CDI.

Statement: There is insufficient evidence to recommend FMT as a treatment for the first episode of CDI.

Moreover, refractory CDI often arises as a severe disease, 27 which may be a life-threatening condition. As FMT appears to be an effective treatment of severe CDI, 2 , 30–32 it can also be considered an effective therapeutic option in CDI that is unresponsive to the usual antimicrobial treatment. However, further studies will need to consolidate this evidence.

Comment: A recent systematic review found seven non-randomised observational studies in which FMT was used to treat patients with refractory CDI. 6 None of them compared FMT with other therapies, and efficacy rates varied widely among studies (0%–100%; overall resolution rate, 55%). In a recent double-bind RCT aimed at comparing frozen FMT with fresh FMT for the treatment of rCDI, a small portion of patients with refractory CDI were successfully treated both with frozen (per-protocol, 100%) and fresh faeces (per-protocol, 67%). 3

Based on this evidence, the panel encourages the use of FMT in clinical practice for the treatment of rCDI. This may also help decrease the associated economic burden on healthcare system, according to recent cost-effectiveness analysis. 10 , 11

The recurrence of CDI may lead to clinical progression of the disease. 27 Therefore, the recurrence of CDI is more likely to appear as a severe disease rather than a first episode and is associated with a higher mortality. 28 , 29 There are several studies that have shown a good efficacy of FMT for severe CDI using a variety of FMT protocols. 2 , 30–32

In all these studies, FMT also showed an excellent safety profile, at least in the short-term follow-up, as only few and almost mild adverse events were reported. Currently, long-term safety data are lacking. In theory, it may be possible to transmit potentially harmful microbiota traits, which may be not apparent for decades. This possibility should however be envisaged in the context of a favourable risk-benefit ratio, as FMT is a highly efficacious treatment for rCDI and can represent a life-saving treatment for affected patients.

Moreover, in several systematic reviews and meta-analyses, including both RCTs and non-controlled studies, rCDI resolution rates achieved by FMT ranged between 85% and 89.7%. 4–6

Comment: Infusion of faecal microbiota from a healthy donor into a recipient individual with CDI can restore the healthy microbial flora in the diseased colon, leading to the resolution of symptoms. In two open-label RCTs including patients with rCDI, FMT showed significantly higher resolution rates than vancomycin (94% and 90% vs 31% and 26%, respectively). 1 , 2 Both studies have a small sample of enrolled subjects, as they were stopped early for futility. The excellent efficacy rates (above 80%) of FMT for the treatment of rCDI confirmed afterwards with a larger double-blind RCT, aimed at comparing frozen FMT with fresh FMT (83.5% and 85.1%, respectively at per-protocol analysis). 3 A further double-blind RCT in which FMT using donor stool (heterologous) was compared with FMT using patient’s own (autologous) stool (this latter study was published after the conclusion of this consensus conference). 26

Statement: FMT is recommended as treatment option for both mild and severe rCDI. Its implementation in clinical practice is recommended.

The experts panel took into account other clinical indications for a possible use of FMT in the clinical practice, such as IBD, IBS, metabolic disorders, paediatrics, but for none of them emerged an evidence-based recommendation to use FMT except that in a context of research (see online supplementary 1 ).

Comment: According to the available evidence, there is no significant difference between patient-selected and anonymous healthy donors in terms of FMT outcomes, at least for rCDI, as shown by a meta-analysis. 4 The use of both related and unrelated donors for FMT provided excellent rates of rCDI resolution in several RCTs and large case series. 1 , 2 , 36 , 37 Nevertheless, randomised studies comparing directly these categories of donors are still lacking. The use of unrelated healthy donors may be useful in FMT centres, with large volume of patients, to satisfy FMT requests. The availability of a stool bank in those centres is of utmost importance. Choice of related or unrelated donors might be driven by specific needs (eg, specific FMT indications, such as when there is a potential advantage of choosing unrelated donor to treat conditions with a pathogenic genetic basis; or in particular research designs). Finally, the advantage of excluding either related (eg, for IBD) or unrelated (eg, for paediatrics) donors still needs additional research to be addressed.

Statement: Related or unrelated donors can be selected when FMT is performed to treat CDI. For other indications, the choice may be driven by specific needs.

Most of the complications reported in the literature are principally related to the route of administration of faecal infusion and not to transmission of infection. However, there is poor evidence suggesting a precise deadline of donor testing before donation.

Comment: The primary purpose of donor testing is to check the donor for infectious diseases potentially transmittable to the recipient. The selected blood and stool exams were shown to achieve an excellent safety profile in several RCTs of FMT. 1–3 , 7 , 14–16 , 36 Some testing should be mandatory, with others as options, according to geographical areas (eg, human T-lymphotropic virus types I and II antibodies, or Strongyloides stercoralis), clinical conditions of recipients (eg, cytomegalovirus IgG, viral capsid antigen IgG, bacterial culture for Vibrio cholera and Listeria monocytogenes, antigens and/or acid fast-staining for Isospora and Microsporidia in the case of immunosuppressed recipients) or medical history of donors (eg, calprotectin). Although for some potential pathogen (such as human T-lymphotropic virus) there is no knowledge on transmission via FMT, the panel has expanded the boundaries of screening tests in order to ensure maximum safety profile to patients. Part of blood testing, moreover, is required by the European Commission for the selection of allogenic living donors of human tissue transplants and by the French National Guidelines of FMT for rCDI. 20 , 35

Statement: Suitable donors for FMT should undergo both blood and stool testing at most 4 weeks before donation (see box 3 ). If there are no changes in donor’s health and specific circumstances, testing may be repeated up to 8 weeks.

An alternative way, without exploiting the questionnaire, is to collect donor stools just after the results of the exams, and freeze and store them, if the centre has expertise in managing frozen material and has the availability of authorised stool bank facilities.

Comment: To ensure the safest possible procedure, the panel recommends a further check of recent history and risk factors of selected and laboratory screened donors on the same day of donation. The above recommendation is supported by excellent safety data from several RCTs, 1 , 2 , 14 , 15 and is also reported in the French National Guidelines of FMT for rCDI. 22

Diarrhoea (more than three loose or liquid stools per day) among members of the entourage (including children) within 4 weeks of donation

Use of antibiotics or other drugs that may impair gut microbiota, new sexual partners or travels abroad since the last screening

Statement: Screened donors have to undergo a further interview on the same day of the donation, in order to check any recently onset potentially harmful issue (see box 2 ).

Comment: The main objective of donor selection is to reduce and prevent any adverse event related to the infused faecal material. Therefore, the panel recommends general exclusion criteria for donor selection before performing FMT, regardless of indications. As a first step, all potential donors should undergo a general questionnaire (preferably written) focused on their medical history and lifestyle habits to identify the risk factors. This is extremely relevant to exclude issues not detectable through blood and stool testing. To reduce the risk of donor comorbidities, individuals aged <60 years should be preferred. However, this indication cannot be mandatory in order not to foreclose the use of intimate healthy partners merely because of age. The panel chose the exclusion criteria shown in box 1 because they are the requirements of the European Commission for the selection of allogenic living donors of human tissue transplants. 35 In addition, further exclusion criteria, especially those related to the presence of GI disorders, or drugs which may impair microbiota, were considered, based on available evidence. Indeed, the use of these or similar exclusion criteria assured, as shown in several RCTs, large case series and systematic review, 1–3 , 5 , 7 , 14–16 , 36 the occurrence of no or few adverse events related to the infusion of donors' material, as also stated in the French National Guidelines of FMT for the treatment of rCDI. 22

Statement: Potential donors for FMT have to undergo, at the beginning of the selection process, a medical interview to exclude history and risk factors (see box 1 ).

On the day of faecal infusion, faecal suspension should be thawed in a warm (37°C) water bath. After thawing, saline solution could be added to obtain a desired suspension volume. The thawed faecal material should be infused within 6 hours after defrosting. Since microbial cells are sensitive after defrosting, 50 repetitive thawing and freezing should be avoided and frozen samples should be prepared in doses needed for each infusion procedure.

The final suspension should be clearly labelled and traceable (using codes similar as for blood and tissue transplantation) and stored at −80°C. Storing at −20°C might be feasible, 3 but should be avoided since it allows some enzymes to be active, which can lead to degradation of sensitive microbial populations (eg, Bacteroidetes). 49 Nevertheless, further studies are needed to show the optimal storage temperature of the frozen samples.

Comment: The FMT approach based on frozen faeces is essential for the development of a stool bank and is the optimal way to standardise the FMT process and to allow the availability of stool on demand, without delays associated with new screening. Comparative RCTs have shown similar efficiency of FMT performed with fresh and frozen faecal samples for the treatment of rCDI. 3 , 38 Similar to fresh samples, preparation of frozen faecal suspensions under normal air or under oxygen-free atmosphere yielded similar resolution rate. 3 , 38 , 45 For indications other than rCDI, anaerobic preparation might be more relevant, as indicated by higher resolution rates achieved with anerobically prepared samples when compared with aerobically prepared faecal suspensions. 3 , 14 The panel recommends to use at least 30 g of donor faeces and 150 mL of saline solution. Before freezing, glycerol should be added up to a final concentration of 10%. 38 Glycerol protects microbial cells from damage induced by freezing. 46 On the contrary, the use of ethanol is still controversial. Despite encouraging phase I findings, 47 ethanol-treated mix of sporulating bacteria failed to treat rCDI in a phase II study. 48 It is plausible that ethanol, beyond killing pathogens, can also severely alter the composition of commensal microbiota, possibly eliminating critical elements such as bacteriophages, fungi and non-sporulating bacterial components.

Statement: Frozen faecal material can be used for FMT. A minimum set of general steps has to be followed for the preparation of frozen material (see box 4 ).

Moreover, the panel recommends faecal material to be diluted with sterile saline solution (0.9%) with three to five times larger volume of solvent (eg, 30 g of faeces to be diluted in 150 mL of saline). In some studies, water has been successfully used as a solvent of faecal material (eg, Satokari et al). 7 , 38 Although it has been reported that the preference should be given to saline as this solvent enables better preservation of microbes, 44 conclusive comparative studies are missing. After homogenisation, the solids should be strained using gauze, 3 tea strainer or similar device 37 and suspension should be poured into a sterile container.

The amount of used faecal material varies from study to study, with most studies using 50 g of faeces 42 ; nevertheless, recently, even 30 g of faecal material were shown to be sufficient for successful FMT. 7 , 38 , 43 However, the panel recognises that stool weight is a highly imperfect measure of microbiota quantity and acknowledge that there is wide variability in microbial content in stool between individuals and even different donations.

Comment: The general steps ( box 4 ) recommended by this statement are based on what has been described, but never rigorously tested. There are no reported studies comparing different preparation protocols of fresh faecal material, but the protocols used in different studies are comparable and allow good/moderate evidence of suitable protocols of fresh faecal preparation for FMT treatment of rCDI. 1–3 , 7 , 38 The treatment efficacy with all considered protocols has been over 80%. When reported, fresh faeces have been used for transplant within 6 hours after evacuation or on the same day. 1–3 , 7 , 38 Both anaerobically and aerobically prepared samples are efficient in treatment of rCDI. 1–3 , 7 , 38 This could be due to the fact that a considerable part of the bacterial genera of healthy microbiota produce resilient spores allowing interindividual transfer of at least a proportion of oxygen-sensitive intestinal bacteria. 39 Given that these spore-forming bacteria typically represent about one-third of gut bacteria, 39 and that disorders associated with microbiota alteration, such as IBS or IBD, are typically defined by lower abundance of anaerobic bacteria, 40 , 41 it is rational to expect that the anaerobic processing of samples would be relevant for FMT success in the treatment of these disorders.

On the day of faecal infusion, faecal suspension should be thawed in a warm (37°C) water bath and infused within 6 hours from thawing

At least 30 g of donor faeces and 150 mL of saline solution should be used

Faecal material should be suspended in saline using a blender or manual effort and sieved in order to avoid the clogging of infusion syringes and tubes

To protect anaerobic bacteria, the storage and preparation should be as brief as possible

Statement: A minimum set of general steps has to be followed for the preparation of fresh faeces (see box 4 ).

Comment: Treatment response implies clinical improvement, 13 as reduction of stool frequency and improvement of stool consistency, as well as amelioration of other parameters of disease severity as laboratory parameters, radiological and/or endoscopic findings. Repetition of C. difficile testing in stool samples is generally not recommended, as toxins might stay positive for weeks. In clinical practice, it is not rational to distinguish between recurrence and relapse of CDI.

Statement: Patients receiving FMT for CDI should be followed up for at least 8 weeks.

Comment: To date, there are not enough long-term follow-up data to estimate potential FMT-related diseases. A single case of weight gain has been reported, as well as the development of peripheral neuropathy, Sjögren’s disease, idiopathic thrombocytopenic purpura and rheumatoid arthritis. 37 , 81 There are other anecdotal reports about improvement of non CDI-related diseases after FMT as IBS, chronic constipation, antibiotic-induced non-infectious colitis as well as extraintestinal diseases as Parkinson's disease, multiple sclerosis and idiopathic thrombocytopenic purpura. 82 , 83 However, causality to FMT remains unclear. The transmission of malignant, autoimmune, metabolic or neuropsychiatric diseases have been discussed and at least partially shown in animal models. 84 Patients with CDI are known to be at high risk of postinfectious IBS. 85 However, the role of FMT in this context has to be elucidated.

Most common short term adverse events after FMT due to CDI have been described as diarrhoea, abdominal cramps, belching, constipation, fever as well as Gram-negative bacteraemia and perforation. 1 , 42 , 65 In IBD, diarrhoea, abdominal bloating and cramping, fever and deterioration of disease have been reported. 51 , 58 , 77 , 78 Mortality has been reported as an outcome due to aspiration during sedation after FMT via colonoscopy. 79 Another patient died after regurgitation of faecal material infused into the duodenum under general anaesthesia. 67 After delivering donor stool via a pre-existing nasogastric tube, another patient developed septic shock and toxic megacolon and died after colonic resection. 80 Finally, a patient had a sever septic shock due to the faecal aspiration following endoscopic peroral jejunal FMT. 54

Comment: Infection control practices of patients with rCDI should be performed as recommended by international guidelines, 12 , 13 according to disease severity and comorbidities. The need for hospitalisation of patients with other underlying diseases depends on the diagnosis and clinical condition. When repeated faecal infusions are necessary, provided that the patient’s condition is good, further applications can be performed in an outpatient setting. 15 The duration of the observation period has not been defined yet, as it depends on the route of delivery, the underlying diseases and the general condition of the patient. 22

Statement: Recipients should be monitored for the occurrence of possible acute complications related to the procedure.

In other indications than rCDI, such as IBD, repeated FMTs have been reported to increase the success rate. 14 , 15 , 51 , 75 , 76 Nevertheless, in many other potential indications than rCDI proven standard procedures are lacking.

Comment: In severe CDI infection, especially in colitis with endoscopic evidence of pseudomembranes, repetition of faecal infusion had been necessary to induce clinical remission. 2 , 30–32 , 70–72 Success of FMT may also depend on the amount of delivered faecal suspension, which can be increased by multiple infusions. 3 Factors negatively influencing the outcome of FMT might be low volume of donor stool, severe colitis and/or ongoing antibiotic treatment and route of delivery, together with frequency of CDI-related hospitalisation. 72–74 Stool from another donor should be considered in case of failed clinical resolution despite multiple infusions.

Comment: FMT has been performed in elderly, 65 immunocompromised 66 , 67 as well as in critically ill patients, via upper and lower GI tract. However, the rate of serious adverse events in recipients who received FMT via upper GI tract seems to be higher. Aspiration pneumonia due to nausea and vomiting by a nasogastric or nasojejunal tube has been reported. 1 , 57 , 68 Fever has been observed both after via-upper GI and (less frequently) after via-colonoscopy infusion. 38 , 55 , 58 , 67 , 68 In the study by Moayyedi et al, 15 rectal abscesses after via-enemas infusion of faeces or placebo were reported in patients with UC. However, perianal disease in UC is also known to occur regardless of FMT. 69

Statement: FMT appears to be safe even in immunocompromised and critically ill patients regardless the route of delivery. In case of critically ill patients, faecal infusion by enema(s) should be preferred.

Protocols using capsules containing faecal preparations for oral administration are up to now under investigation, with mixed results. 62–64 This approach could expand the availability of FMT therapy in terms of accessibility and patient desirability. However, conclusive results are lacking, as comparative trials have not been performed yet.

Due to side effects and gut microbiota modifying effects of prokinetics and PPIs, their use should be considered cautiously. 60 , 61

The peri-interventional use of prokinetics and proton pump inhibitors (PPIs) has been reported in case of upper GI delivery. 20 , 58 , 59 However, their beneficial use has not been proven yet.

Comment: When applied to the upper GI tract, total volumes of inserted bacterial suspension are much lower, about 25–50 mL compared with lower GI tract application. 57 Volumes up to 500 mL have been placed via nasojejunal tubes over a longer period (2–3 min/50 mL). 1

Statement: FMT can be performed via upper GI tract. The faecal suspension can be delivered through the working channel of a gastroscope, or through nasogastric, nasojejunal or gastrostomy tube. Patients must be kept in a 45° upright position for 4 hours after infusion in order to prevent aspiration.

Comment: Enema as route of administration have been reported in rCDI, in paediatric patients as well as in critical ill patients when colonoscopy was contraindicated. 3–5 In other indications than rCDI, some study protocols implicate repeated enemas to increase clinical success. 15 The enema route of administration may offer some advantages, as it is widely available and does not require costly devices. It is less invasive than other routes.

Statement: FMT can be applied by enema. Patients should be instructed to hold the infused material for at least 30 min and to remain supine to minimise the urge to defecate. The procedure could be repeated. 3

Comment: Different routes of faecal delivery have been reported. Many systematic reviews and meta-analyses reported that colonoscopy achieved higher resolution rates of rCDI and similar safety profile than other routes of delivery. 4–6 Depending on the general condition of the recipient, donor stool can be applied to the right or to the left colon. 2 , 7 , 26 , 31 , 55 It does not seem to be very essential to dispose the suspension in multiple, smaller portions as it will distribute by gut peristalsis anyway. An amount of 200–500 mL of faecal suspension obtained from 20 to 100 g faeces can be delivered safely to the colon in contrast to other sites. 2 , 26 , 37 , 45 , 56

Statement: When possible, donor stools should be infused into the right colon via the working channel of the colonoscope. In cases of severe colitis, faecal suspension can be disposed in the left colon for safety reasons.

Comment: For delivery by colonoscopy, recipients receive a conventional colon lavage routinely given prior to colonoscopic examination. 2 , 17 , 20 , 21 As bowel lavage is able to reduce the abundance of C. difficile, it is also reasonable to suggest it also when FMT is performed via upper GI tract. 1 , 2 , 52 , 53 However, there is some report showing high success rate of FMT performed via upper GI tract and without prior bowel lavage. 54 For delivery via enema, no preparation instructions have been developed. In particular, with a large RCT to treat rCDI by FMT, Lee et al obtained high cure rates using enema faecal delivery without colon lavage prior the procedure. 3

Statement: Recipients should be prepared with bowel lavage by polyethylene glycol before procedure when FMT is performed by upper route or by colonoscopy.

In indications other than rCDI, for example, IBD, there are no high quality data to support any recommendation, even though antibiotic pretreatment could potentially increase clinical response. 51

Comment: In patients with rCDI preinfusion use of metronidazole, vancomycin or fidaxomicin should be performed for at least 3 days, 1–3 , 17 , 20–22 in order to repress the abundance of intestinal C. difficile. However, antibiotics should be stopped 12–48 hours before stool application.

Statement: Patients with rCDI should be treated with vancomycin or fidaxomicin at least for 3 days before FMT. Antibiotics should be stopped 12–48 hours before faecal infusion.

Basic requirements for implementing an FMT centre

Key issue: clinical requirements and facilities

Statement: Development of referral FMT centres for the treatment of CDI in clinical practice is encouraged. Centres should be implemented in hospitals with appropriate expertise and facilities.

Quality of evidence: moderate.

Strength of recommendation: strong.

Comment: CDI is the most common cause of diarrhoea in inpatients, and is increasing in incidence, morbidity, mortality and likelihood to recur.8 On the other hand, CDI not responding to standard therapy is a complex disease requiring clinical expertise. FMT achieved significant advantage over standard antibiotic therapy for the management of rCDI in RCTs,1 ,2 and is clearly recognised as a reliable and cost-effective treatment for this burdensome disease.6 ,9 ,11 Therefore, the dissemination of the FMT procedure and the establishment of FMT services into clinical setting could be useful practices to reduce the healthcare burden of rCDI,43 as advocated by physicians interested into gastroenterology and infectious diseases.18 ,19 ,86–89 The development of an FMT centre service could fill the therapeutic gap for the management of the life-threatening rCDI by ensuring the optimal standardisation of the FMT process.

Statement: In order to establish an FMT centre, actively involved members are encouraged to undergo a specific training on FMT processes.

Quality of evidence: low.

Strength of recommendation: strong.

Comment: As with other technical procedures, a specific training period is required before performing FMT. Therefore, physicians who are going to implement FMT in their centre are encouraged to undergo a prior training. Essentially, training on FMT is composed of three parts: clinical training, which includes donor and patient selection, and patient management after FMT; delivery training, which includes the learning of different routes of delivery (eg, enema, colonoscopy, nasoduodenal/nasojejunal tube, etc) and microbiological training (preparation of fresh and frozen faecal material).

Statement: The assemblage of a multidisciplinary team, including gastroenterologists, microbiologists and infectious disease physicians, is encouraged to build an FMT centre.

Quality of evidence: low.

Strength of recommendation: strong.

Comment: There are no published data supporting this statement, and evidence comes from opinions of the panel, based on clinical experience, descriptive case studies or reports of expert committees. Nevertheless, the panel strongly believes that the development of a multidisciplinary team, composed of gastroenterologists, microbiologists and infectious disease physicians, will be extremely useful to build an FMT referral centre, because it would gather different and complementary expertises.

Statement: The availability of several facilities (including endoscopy service, clinical ward and outpatient clinic) is mandatory to implement an FMT centre. FMT can be performed within either inpatient or outpatient setting.

Quality of evidence: low.

Strength of recommendation: strong.

Comment: There are no published data supporting this statement and evidence comes from opinions of the expert panel, based on clinical experience, descriptive case studies or reports of expert committees. Provided that the patient's condition is stable, faecal infusion can be performed in an outpatient setting.20 The panel strongly believes that several facilities are needed to build an FMT referral centre, to offer FMT to both patients with mild and severe CDI, to manage potential adverse events related either to the procedure or to the infused material and to easily follow-up patients after FMT.

Statement: A clinical governance dealing with administrative issues of FMT is recommended to develop an FMT centre.

Quality of evidence: low.

Strength of recommendation: strong.

Comment: There are no published data supporting this statement and evidence comes from opinions of the expert panel, based on clinical experience, descriptive case studies or reports of expert committees. The panel believes that the implementation of a clinical governance within the FMT centre would be extremely useful to deal with administrative issues related to FMT (eg, reimbursements, authorisations), and to overcome all bureaucratical and organisational barriers that may prevent the development and the work of the centre itself.

Key issue: microbiological requirements and facilities

Statement: FMT centres need to have an access or be part of the facility that allows safe processing of human samples (biosafety level 2) including aliquoting, storage and preparation of faeces. Stool banking is encouraged.

Quality of evidence: low.

Strength of recommendation: strong.

Comment: There are no evidence-based data supporting the recommendation of high-quality microbiological requirements for building an FMT Centre. Current recommendations include expert opinions from the expert panel and existing general recommendation including guidelines for safe work practice.90–92

One of the key functions of the FMT centre is the management of blood and faecal samples from either donor or recipients. C. difficile is a pathogen with a biosafety level 2 (cabinet with high-efficiency particulate air filter) and according, safety requirements and recommendation need to be implemented.90 Material processing (fresh faeces, bank of frozen and stored faeces) and safety precautions have to adhere to basic principles for safe preparation of human material, including: rigorous protocols in securing the materials; maintenance of standard operating procedures (SOPs) for the processing; use of certified laboratory testing; definition of quality control tests and standards for the release of the final product.

There are numerous advantages associated with the implementation of stool banks (such as the OpenBiome initiative in the USA). A frozen stool bank allows faecal donors to be recruited and thoroughly screened ahead of time, in a methodical manner, without time pressure and with the potential advantage to reduce the costs associated with donor screening as one donor can serve for multiple FMT donations. Furthermore, the accumulation of expertise and the standardisation of FMT processing ultimately allow optimising the quality control, a greater potential for rigour of testing and the coordination of research activities/outcomes monitoring. Finally, stool banks may improve the accessibility of FMT to centres that otherwise would be unable to provide the service due to inadequate resources to conduct donor recruitment/screening and FMT processing.

Statement: FMT procedure and donor screening documentation should be stored for at least 10 years (unless local requirements consider longer retention period) in order to archive donor material in case of future adverse events.

Quality of evidence: low.

Strength of recommendation: strong.

Comment: Management of the records related to the FMT procedure should be regulated by local health organisation. FMT procedure and donors' and recipients' records should be stored for at least 10 years. This may differ according to local requirements, and longer storage time could be needed. The records of the FMT centre will provide access to the long-term safety data.

Samples of donor faeces before the processing and before the administration to recipients should be stored for a possible microbiological evaluation and qualitative and quantitative characterisation due to safety reasons. The frozen samples should be stored in adequate facilities (or, when specifically regulated, by authorised facilities) and should be clearly labelled with the code of the donor and the date of donation.

Key issue: regulatory requirements

Statement: Appropriate FMT registries should be implemented, in order to collect data concerning indications, procedure, effectiveness and safety profiles.

Quality of evidence: low.

Strength of recommendation: strong.

Comment: There are no published data supporting this statement, and evidence comes from opinions of the expert panel, based on clinical experience or on reports of expert committees. Recently, the American Gastroenterological Association established a National Institutes of Health-funded registry to track patient outcomes associated with FMT,93 with the ultimate objective to protect the well-being of patients and guide further scientific exploration in the field. Such a registry would serve as a source of short-term and long-term information about the clinical practice of FMT in the USA.

FMT has not undergone traditional regulatory approval process of pharmaceutical products with sequential testing leading to large phase III trials assessing efficacy and safety prior to clinical utilisation. Therefore, the panel believes that the creation of registries to be kept at the local (hospital) and/or at the regional, national or international competent authorities is relevant to collect data and could be useful to deal with outcomes and safety issues related to FMT. In order to trace causality of FMT and newly developed diseases, traceability by keeping appropriate registries would be a wise process to learn about potential long-term side effects.

Statement: Specific national rules for the classification of FMT should be followed to implement an FMT centre.

Quality of evidence: low.

Strength of recommendation: strong.

Comment: A commonly acknowledged regulatory classification for FMT has not been established yet in Europe. Several countries have introduced some national rules and others require to be compliant to the European directive 2004/23 on quality and safety of tissues and cells.94

As FMT falls under the category of ‘Substance of Human Origin’ like cells, tissues, milk, etc., the most important requirements should be followed, as for example an adequate facility.

Activities and responsibilities for processing and testing the raw material, for use of equipment, for preservation and storage and for release and distribution of FMT, should be described in SOPs.