In 2015 in the United States, the costs of Clostridium difficile infection (CDI) and associated care were US$6.3 billion.Reference Ramai, Zakhia, Ofosu, Ofori and Reddy1 In Europe, these costs were estimated at €3 billion per year, ranging from €5,798 ($6,448) to €11,202 ($12,457)per episode.Reference Reigadas Ramírez and Bouza2 These costs are expected to nearly double over the next 40 years, depending on the increase of patients at risk of developing CDI. This potentially substantial increase may be explained by the number of elderly patients, antibiotic consumption, the development of resistance to conventional antibiotics (eg, metronidazole, vancomycin, and fidaxomicin), and the emergence of hypervirulent strains.Reference Reigadas Ramírez and Bouza2–Reference Mathias, Curti, Montana, Bornet and Vanelle4 The major problem with CDI is that despite adequate antibiotic therapy, 10%–30% of patients will experience a recurrence, with the risk approaching 60% after the third episode.Reference Cohen, Gerding and Johnson5
Fecal microbiota transplantation (FMT) is an effective therapy for CDI, leading to a significant reduction of recurrent Clostridium difficile infection (rCDI) and reduced incidence of adverse events when compared to conventional antibiotics (eg, vancomycin and fidaxomicin).Reference Mathias, Curti, Montana, Bornet and Vanelle4,Reference Surawicz, Brandt and Binion6–Reference van Beurden, de Groot, van Nood, Nieuwdorp, Keller and Goorhuis8 FMT is currently recommended by both European and American guidelines for multiple rCDI.Reference Debast, Bauer and Kuijper9,Reference McDonald, Gerding and Johnson10 Other indications of FMT are being studied, such as inflammatory bowel diseases (IBDs; eg, ulcerative colitis and Crohn’s disease), obesity, type 2 diabetes, or graft-versus-host disease.Reference Cammarota, Ianiro and Tilg11–Reference Moutinho, Baima and Rigo23 However, to date, FMT has not yet been recommended for use in daily practice for these diseases.
FMT is an intervention in which a fecal suspension from a healthy donor is transferred into the gut of a patient to replace depleted components of the gut microbiota.Reference Gupta, Allen-Vercoe and Petrof24 No general agreement has been reached on the best approach for delivering fecal microbiota nor on the optimal transplantation volume. Thus, various routes of administration have been described either via the upper or via the lower gastrointestinal (GI) tract. FMT via upper GI delivery methods include nasogastric tubes (FMTng) or duodenal infusion (FMTdi), whereas FMT via lower GI delivery methods mainly involve colonoscopy (FMTc) or enema (FMTen).Reference van Nood, Vrieze and Nieuwdorp25 More recently, randomized clinical trials have indicated that the use of oral capsules was not inferior to other modes of administration.Reference Youngster, Mahabamunuge and Systrom26–Reference Kao, Roach and Silva29
The status of FMT is different between countries. The Food and Drug Administration (FDA) and Health Canada consider FMT a biological tissue; the American FDA considers it an investigational new drug; and the European Medicine Agency allows each country the possibility of assigning a qualification. In France, FMT is considered a drug.30–Reference Megerlin and Fouassier32
In the context of rational decision making in health care, and as an integral part of health technology assessment (HTA), the economic evaluation (EE) of healthcare products has become a necessity. One of the major challenges in healthcare research is to provide cost-effectiveness data that are relevant to daily practices and that may be required to optimize the consumption of healthcare resources. Decision making for coverage and reimbursement of new drugs is being increasingly supported by EE in many countries including Australia, Canada, and the United Kingdom.Reference Greenberg, Earle, Fang, Eldar-Lissai and Neumann33,Reference Frederix, Severens, Hövels, Raaijmakers and Schellens34 The quality of EE must be high to instill trust in the accuracy of the results and to support informed decisions. Among several EE methods,35 the most common and recommended in HTA are cost-effectiveness analyses (CEAs) and cost-utility analyses (CUAs).36 The choice of method depends on the nature of the expected health effects of the interventions under study. A CEA is required when health-related quality of life (HRQOL) is not identified as a relevant health effect of the studied interventions; health outcome is measured by the length of life in life years (LYs). Otherwise, CUA is the preferred method when HRQOL is identified as an important health effect of interventions; health outcome is measured by the length of life weighted by a valuation of the HRQOL, represented by health-state utility values (HSUVs), to produce quality-adjusted LYs (QALYs). The HSUV is measured on a scale anchored by 1 (best imaginable health state, ie, perfect health) and 0 (worst imaginable health state, ie, death) using patient preference-based measures.Reference Whitehead and Ali37
In this context, the aims of the present study were to systematically identify and review published FMT-related EEs and to assess their quality.
Method
We adopted an optimal method to enhance the quality of our systematic review based on modified AMSTAR, a reliable and valid measurement tool consisting of 11 items.Reference Shea, Grimshaw and Wells38
Search strategy
A systematic literature search was conducted in PubMed and Cochrane to identify published EEs. An articles was included if it concerned FMT (whatever the disease and its line of treatment), if it reported full or partial economic evaluations, and if it was written in English. An articles was excluded if it did not concern FMT, if it did not report an EE, or if it was a systematic review, an editorial, a comment, a letter to the editor, a practice point, or a poster.
Medical subject headings (MeSH) terms were individually selected using the National Library of Medicine controlled vocabulary thesaurus used for indexing articles for PubMed, (ie, MeSH database) before being combined: (1) cost–benefit analysis or economic evaluation or cost effectiveness or cost analysis or cost savings and (2) fecal microbiota transplantation or fecal transplantation or fecal transplant). The full request in both electronic sources PubMed and Cochrane is presented in Appendix A (online).
Article selection
The titles and abstracts of all identified articles were screened and analyzed by 1 reviewer (S.D.) and checked by a second reviewer (V.N.) to determine whether each article corresponded to the previously defined topic and whether it was an original EE. Second, after initial screening, the full text of each selected article was independently analyzed by both reviewers. In addition, the reference lists of all selected studies were screened to identify other potentially relevant articles that had not been identified by the 2 electronic means. Consensus was reached in cases of disagreement between the 2 reviewers. The main reason for excluding a given article was recorded.
Data extraction
Based on the CHEERS checklist, the following data were identified and extracted from each selected article: year of publication; journal; main location of the first author; sponsor; conflict of interest; topic (ie, CDI or IBD); line of treatment (ie, initial or recurrent for CDI); aim(s) of the EE; examination of both the costs and the consequences of alternative interventions; examination and comparison of at least 2 interventions; type of EE (full, including CEA, cost-minimization analysis, CUA, cost–benefit analysis; or partial, including cost description [ie, cost-of-illness], cost analysis and cost–outcome description); number of interventions compared ≥ 2; compared interventions; population analyzed (hypothetical cohort of patients, patients included in a clinical trial, patients of a national or a local institution); perspective (ie, the point of view from which the costs and health effects are recorded and assessed: societal, healthcare system); country and continent of origin; time horizon (ie, the period during which patients were followed measuring costs and health effects, long enough to reflect all expected differences between the interventions being compared); main data source of costs (derived from local databases(s), national database(s), the literature, etc); year of reference for costs; main data source of effectiveness (derived from the results of clinical trial(s), literature, original study), and especially health-state utility values if applicable; decision making analysis model (yes or no); Markov model in medical decision-making (yes or no); discounting of costs and effectiveness (ie, the reflection of the present value of future costs and health effects) and discount rate; results in terms of costs; effectiveness (LY, QALY, etc) and cost-effectiveness (ICER) or cost-utility (ICUR); and difference between ICER and ICUR according to the threshold and deterministic and probabilistic sensitivity analyses (ie, the characterization of uncertainty; yes [complete], yes [partial], no, and not assessable).
A dominant strategy is both less expensive and more effective than the comparator strategy, whatever the willingness-to-pay threshold. On the contrary, a strategy that is both more expensive and less effective is classified as “dominated.” After ICER analysis, if the strategy is most effective but more costly than the comparator strategy, dominance can be extended to specific willingness-to-pay threshold. Extended dominance rules out any strategy with a higher ICER than a more effective strategy.
To allow direct comparisons across countries, all costs were converted to US dollars and then inflated to the reference year of 2019.Reference Husereau, Drummond and Petrou39,40
Quality assessment of the economic evaluations
A critical appraisal of the methodological quality of health economic evaluations is not simple to carry out, and it may depend on the subjective performance of the evaluator. As previously described, using the Drummond checklist as a scoring system has the advantage of allowing a comparison between different EEs and their results; thus, it is more objective and provides a way to measure the transferability of an EE and its results.Reference Husereau, Drummond and Petrou39,Reference Drummond, Sculpher, Claxton, Stoddart and Torrance41,Reference Nerich, Saing and Gamper42 However, as with any rating scale, a major limitation of the Drummond checklist is the likelihood of bias, even when the interobserver variability between reviewers is good.Reference Gonzalez-Perez43,Reference Lapointe-Shaw, Tran and Coyte44 Thus, the quality assessment of each selected CUA was performed by 2 independent reviewers (S.D. and V.N.). When there was disagreement, a consensus was reached using this checklist based on 10 questions; the checklist provided a framework for assessing methodological quality. For each question, there were 4 possible responses: yes, no, not clear, and not appropriate. One point was assigned for each “yes” response. The lowest and highest possible scores were thus 0 and 10, respectively. A score ≥7 was considered high quality.Reference Drummond, Sculpher, Claxton, Stoddart and Torrance41
Results
Article selection
The search in both PubMed and Cochrane identified 46 articles, but 1 was a duplicate. In addition, 36 were excluded after screening and analyzing titles and abstracts for exclusion criteria (Fig. 1). In total, 9 articles were analyzed and eligible for the systematic review (Appendix B online).Reference Lapointe-Shaw, Tran and Coyte44–Reference Jiang, Leung, Ip and You52
Fig. 1. CONSORT flow diagram of literature review. (A) not only FMT. (B) FMT but did not report economic evaluation. (C) Systematic review, editorial, comment, letter to the editor, practice point and poster. Note. EEs, economic evaluations; FMT, fecal microbiota transplantation.
Synthesis of basic elements of economic evaluations
The characteristics of the 9 EEs are summarized in Table 1. All EEs were published in a clinical journal after 2013. Only 1 article evaluated initial CDI (iCDI), but 7 evaluated rCDI. Only 1 article assessed FMT EE in IBD (both ulcerative colitis and Crohn’s disease) and 8 EEs were full EEs. At least 2 interventions were assessed by all authors. Authors usually conducted their analyses from the perspective of the healthcare system (n = 6). The main continent of origin was North America (n = 5), and the time horizon ranged from 1 episode (n = 1) to 1 year (n = 7).
Table 1. Characteristics of the 9 Economic Evaluations Included in the Systematic Review
A synthesis of the basic elements of the 9 EEs is summarized in Table 2. Cost was derived from local or national database(s), the literature, or assumptions. Effectiveness was essentially derived from the literature. A model had been developed in 7 EEs. A discounting of costs and effectiveness was carried out in 25% of the cases. A complete sensitivity analysis had been conducted in 3 EEs.
Table 2. Synthesis of Basic Elements of the 9 Economic Evaluations Included in the Systematic Review
Synthesis of results of economic evaluation
In the treatment of iCDI, FMTc, FMTen and FMTng were dominant versus vancomycin (V) but were dominated versus metronidazole (M) (Table 3).Reference Varier, Biltaji and Smith51 In the treatment of rCDI, FMTc, FMTen, FMTdi and FMTng were dominant versus vancomycin, metronidazole, and fidaxomicin in 6 EEsReference Lapointe-Shaw, Tran and Coyte44,Reference Baro, Galperine and Denies46–Reference Konijeti, Sauk, Shrime, Gupta and Ananthakrishnan49,Reference Jiang, Leung, Ip and You52 and were cost-effective at a threshold of ≤$50,000/QALY for Merlo et alReference Merlo, Graves, Brain and Connelly48 and Baro et al.Reference Baro, Galperine and Denies46 In the treatment of IBD, FMT was cost-effective at the willingness-to-pay threshold of $20,158. Among different modes of administration (ie, FMTc, FMTen, FMTdi, and FMTng), FMTc was dominant.Reference Lapointe-Shaw, Tran and Coyte44,Reference Baro, Galperine and Denies46
TABLE 3. Synthesis of Characteristics and Results of the 9 Economic Evaluations Included in the Systematic Review
Quality assessment of the economic evaluations
The median Drummond score of the 9 selected EEs was 7 (range, 6–10; first quartile, 6; third quartile, 9). The result of the quality assessment is provided in Fig. 2 and Appendix C (online). Of these 9 EEs, 6 had a Drummond score ≥ 7. Each was a full EE.
Fig. 2. Percentage of ‘‘Yes’’ for each question of Drummond 10-point checklist for assessing economic evaluations.
Discussion
In this systematic review, we aimed to assess the economic impact of FMT as a new therapy. A trustworthy and valid measurement tool, AMSTAR, was used first to search for and select articles and then to extract data to optimize the quality of this systematic review.Reference Shea, Grimshaw and Wells38 In addition, 9 FMT-related EEs were identified and were deemed cost-effective for rCDI but not for iCDI. A single EE found FMT to be cost-effective for IBD. According to the only EE about IBD, the treatment was cost-effective, but more EEs are necessary to confirm this finding. To date, American and European guidelines only recommend FMT for multiple rCDI.Reference Surawicz, Brandt and Binion6,Reference van Beurden, de Groot, van Nood, Nieuwdorp, Keller and Goorhuis8,Reference Cammarota, Ianiro and Tilg11 An EE showed that the strategy of American guidelines recommending the use of FMT in multiple rCDI was a cost-effective one, thus validating its usefulness.Reference Rajasingham, Enns, Khoruts and Vaughn53 Although FMT seems to be a cost-effective and promising therapy for rCDI and other diseases like IBD, it will be necessary to provide decision makers of healthcare policies with additional EEs to adequately inform the adoption of appropriate guidelines.
The 9 selected FMT-related EEs are full and 8 of them are CUAs. In infectious diseases, a CUA should be preferred to a CEA, especially for the EE of FMT, because of its special impact on the quality of life. To improve the quality of future EEs, it is important to provide a thorough evaluation, which has not been done thus far.Reference Debast, Bauer and Kuijper9 Thus, using disease-related health-state utility values is necessary to precisely measure the length of life weighted by a valuation of the HRQOL. To date, no CDI- or IBD-related health-state utility values are available for FMT. Therefore, the estimation of quality of life and cost-utility may be biased.
Methods used by authors and topics addressed by EEs varied among the 9 EEs. In fact, the lines of treatment could be different: initial, greater than or equal to first recurrence, greater than or equal to second recurrence, or greater than or equal to third recurrence for CDI. Dosages and comparative reference treatments (eg, vancomycin, fidaxomicin, and metronidazole) were also disparate among the EEs.Reference Lapointe-Shaw, Tran and Coyte44,Reference Varier, Biltaji and Smith47,Reference Konijeti, Sauk, Shrime, Gupta and Ananthakrishnan49 The various routes of FMT administration were compared: FMTc, FMTen, FMTdi, and FMTng. Analyzed population, perspective, time horizon, source of costs, and source of effectiveness were also different among the 9 EEs. In addition, 8 EEs were based on a model such as the Markov model.
Overall, FMT interventions generate many costs. All of the expenses must be considered to estimate the exact cost of this therapy: donor selection, preparation of fecal material with antibiotics, fecal delivery, and both short-term and long-term monitoring of patients for efficacy outcomes and adverse events.Reference Ding, Li and Li20,Reference Sokol, Galperine and Kapel54 There are differences among countries, especially regarding donor selection. Moreover, FMT has some risks such as the transmission of multidrug-resistant organisms. The authors of various studies did not calculate the overall cost of FMT in the same way, which influenced the results of their EEs. For example, Merlo et alReference Merlo, Graves, Brain and Connelly48 considered that the overall cost included blood and stool screening, human, material and technical time, pretreatment, obtaining, storing, preparing and administering the fecal infusion, whereas Varier et alReference Varier, Biltaji and Smith47 considered that it included only screening donors and recipients, and the cost of antibiotics and adverse events.
In the treatment of rCDI, FMT was cost-effective at a threshold of ≤$50,000/QALY for Merlo et alReference Merlo, Graves, Brain and Connelly48 and Baro et al.Reference Baro, Galperine and Denies46 However, FMT was also cost-effective because it was dominant in 4 EEs compared with vancomycin and was dominant in 2 EEs compared with fidaxomicin.Reference Lapointe-Shaw, Tran and Coyte44,Reference Baro, Galperine and Denies46–Reference Konijeti, Sauk, Shrime, Gupta and Ananthakrishnan49,Reference Jiang, Leung, Ip and You52 In the treatment of iCDI, FMT was dominated by metronidazole but was dominant versus vancomycin.Reference Varier, Biltaji and Smith51 Nevertheless, metronidazole is no longer recommended as a first-line therapy in this indication. Only 1 EE regarding IBD included and demonstrated its efficiency at the willingness-to-pay threshold of $20,158. The use of drugs was significantly reduced after FMT in this study.Reference Lapointe-Shaw, Tran and Coyte44
All CUAs related to rCDI and IBD in this systematic review concluded that FMT was cost-effective in comparison with other reference treatments at a threshold of ≤$50,000/QALY.Reference Lapointe-Shaw, Tran and Coyte44–Reference Jiang, Leung, Ip and You52 Another review that included only 4 studies about FMT in rCDI also concluded that FMT was cost-effective.Reference Le, Nghiem, Mullen and Deshpande55 FMT was considered cost-effective if it was less expensive and more effective than other treatments (FMT was dominant).Reference Lapointe-Shaw, Tran and Coyte44,Reference Varier, Biltaji and Smith47,Reference Merlo, Graves, Brain and Connelly48,Reference Jiang, Leung, Ip and You52 Otherwise, FMT was more expensive but more effective compared with other treatments.Reference Baro, Galperine and Denies46,Reference Merlo, Graves, Brain and Connelly48 Moreover, in these 4 studies, different routes of fecal delivery were reported, and FMTc achieved higher resolution rates. Indeed, FMTen, FMTdi, and FMTng had a QALY lower than FMTc. FMTc was dominant compared with other modes of administration.Reference Lapointe-Shaw, Tran and Coyte44,Reference Baro, Galperine and Denies46
Our analysis has several limitations other than those previously identified. First, it included only EEs written in English. Secondly, unpublished EEs (grey literature) and EEs of other electronic sources were not included, which may also have introduced some bias. Thirdly, we may have overestimated or underestimated the methodological quality of these EEs. Indeed, assessing methodological quality was often difficult and could have involved some subjectivity on the part of the reviewers.
In conclusion, after examining 9 EEs, FMT seems to be a promising and cost-effective treatment for rCDI but not for iCDI. American and European guidelines recommend the use of FMT for the treatment of rCDI, but additional EEs are needed to accurately identify its place within the therapeutic armamentarium of this disease. In addition, more EE studies on other diseases like IBD are necessary to assess FMT efficiency for new indications. Despite the limited number of EEs in the present study, our review suggests that FMT may be a cost-saving intervention in managing rCDI. It also provides decision makers of healthcare policies with additional information about the cost-effectiveness of FMT.
Supplementary material
To view supplementary material for this article, please visit https://doi.org/10.1017/ice.2019.371
Acknowledgments
None.
Financial support
No financial support was provided relevant to this article.
Conflicts of interest
All authors report no conflicts of interest relevant to this article.
