A Systematic Review of the Cost-Effectiveness of Biologics for the Treatment of Inflammatory Bowel Diseases

Background Biologics are used for the treatment of inflammatory bowel diseases, Crohn´s disease and ulcerative colitis refractory to conventional treatment. In order to allocate healthcare spending efficiently, costly biologics for inflammatory bowel diseases are an important target for cost-effectiveness analyses. The aim of this study was to systemically review all published literature on the cost-effectiveness of biologics for inflammatory bowel diseases and to evaluate the methodological quality of cost-effectiveness analyses. Methods A literature search was performed using Medline (Ovid), Cochrane Library, and SCOPUS. All cost-utility analyses comparing biologics with conventional medical treatment, another biologic treatment, placebo, or surgery for the treatment of inflammatory bowel diseases in adults were included in this review. All costs were converted to the 2014 euro. The methodological quality of the included studies was assessed by Drummond’s, Philips’, and the Consolidated Health Economic Evaluation Reporting Standards checklist. Results Altogether, 25 studies were included in the review. Among the patients refractory to conventional medical treatment, the incremental cost-effectiveness ratio ranged from dominance to 549,335 €/Quality-Adjusted Life Year compared to the incremental cost-effectiveness ratio associated with conventional medical treatment. When comparing biologics with another biologic treatment, the incremental cost-effectiveness ratio ranged from dominance to 24,012,483 €/Quality-Adjusted Life Year. A study including both direct and indirect costs produced more favorable incremental cost-effectiveness ratios than those produced by studies including only direct costs. Conclusions With a threshold of 35,000 €/Quality-Adjusted Life Year, biologics seem to be cost-effective for the induction treatment of active and severe inflammatory bowel disease. Between biologics, the cost-effectiveness remains unclear.


Introduction
. Direct costs denote the resources consumed, while indirect costs are costs due to the loss of productivity related to illness or death. CEA can be conducted using an empirical, observational, or modeling approach [24]. The modeling study appears to be the most common type of CEA, combining clinical data and cost data from many sources. Modeling studies can be tested by sensitivity analysis.
The CEAs provide valuable information for health care decision-makers and enable efficient spending [24]. The aim of this systematic review is to evaluate existing relevant evidence regarding the cost-effectiveness of biologics for the treatment of IBDs. The cost-effectiveness of biologics is compared with placebo treatment, conventional medical treatment, surgery, and another biologic treatment for adults with diagnosed IBD. The aim of this review is also to analyze the source of effectiveness of CEAs. Furthermore, this review assesses the quality of the included CEAs using three different quality assessment checklists.

Literature Search
A comprehensive literature search on the cost-effectiveness of biologics for the treatment of IBDs was performed using Medline (Ovid), Cochrane Library (Cochrane Database of Systematic Reviews, Database of Abstracts of Reviews of Effects, Cochrane Central Register of Controlled Trials, Cochrane Methodology Register, Heath Technology Assessment Database, and NHS Economic Evaluation Database), and SCOPUS (including Embase) in June 2014. The search strategies were developed together with an information specialist. The reference lists of relevant articles were scrutinized. Furthermore, the grey literature and other relevant websites and databases (Centre for Reviews and Dissemination, Current Controlled Trials, Clinical Trials.gov, and PROSPERO) were hand-searched for relevant studies.
The electronic search strategy was based on patients (IBD, CD, or UC), intervention (biologics), and outcomes (ICER) in different spellings (S1 File). The biologics granted a marketing authorization by the European Medicines Agency (EMA) or US Food and Drug Administration (FDA) before May 2014 were included in the literature search strategy [9,10]. No restriction was set based on the year of the publication.

Study Selection
The study selection was based on the inclusion and the exclusion criteria formulated by the framework of PICOTS i.e., population, intervention, comparator, outcome, timing, and setting (S1 Table) [27]. The study selection procedure encompassed three main stages. At the first stage, hits from the electronic databases were imported into reference management software (RefWorks). After removing duplicate citations, the second stage focused on the evaluation of the remaining studies based on their titles and abstracts. Studies clearly indicated as irrelevant to the study subject were excluded. The full articles retrieved that met the inclusion criteria are included in the current review. The identified abstracts and full texts were screened for eligibility by one reviewer (SH) and the second reviewer (MB) was consulted.

Data Extraction
Our data extraction form was based on the Cochrane Handbook for Systematic Reviews of Intervention and the abstract form of the NHS Economic Evaluation Database [28,29]. The following items were extracted: patients, interventions, controls, study design (the type of economic evaluation and modeling, perspective, time horizon, country, included costs, the methods of measuring and valuing outcomes and benefits, discount rate, currency, price year, and the type of sensitivity analysis) and outcomes (total costs and benefits, ICER, and the results of sensitivity analysis). In order to facilitate the comparison of estimates collected from different studies, all costs were converted to 2014 euro using the exchange rates of the European Central Bank and the value of money index published by Statistics Finland [30,31]. Data were extracted using Microsoft Excel and performed by one assessor (SH) and ambiguities were solved by another assessor (MB) for accuracy.

Quality Assessment
The methodological quality of the studies was assessed using three standardized checklists. All studies were assessed using Drummond's checklist, published by the British Medical Journal Working Party, and the Consolidated Health Economic Evaluation Reporting Standards (CHEERS) guidelines [32,33]. In addition, economic evaluations using modeling methods were assessed using Philips' checklist [34]. The quality assessment was conducted by one assessor (SH) and ambiguities were resolved by consulting another assessor (MB).

Synthesizing Data
The results of the included CUAs were stratified into 4 subgroups by the type of previous treatments: 1) the cost-effectiveness of biologics in patients without previous treatment, 2) the costeffectiveness of biologics in patients with previous conventional medical treatment, 3) the costeffectiveness of biologics in patients with previous surgery, and 4) the cost-effectiveness of biologics in patients with previous biologic treatment. Biologic treatments were stratified under three dosing regimens: a single dose, an episodic treatment, or a maintenance treatment. ICERs were presented as principal outcomes. In this study, we analyzed the cost-effectiveness of biologics using the willingness-to-pay threshold of 35,000 €/QALY. A quantitative synthesis of the study results was not possible because of the heterogeneity in participants, interventions and study designs.

Literature Search
The database search identified 1828 references, of which 461 were removed as duplicates, leaving 50 studies to be screened by abstracts and titles for further evaluation. After the assessment of the full text, 31 studies were excluded (S2 File) and 19 studies were included in the review. Additionally, six full-text articles were included, of which two were found from the bibliographies of already included studies [35,36] and four from the structured abstracts identified by the literature search [19,22,37,38]. The hand search revealed no further publications. Altogether, 25 studies were included in the review [19,21,22,. Study selection is presented in a flow diagram in Fig 1.

Cost-Effectiveness of Biologics in Patients with No Previous Treatment
In two studies, the cost-effectiveness of biologics was evaluated in CD patients with no previous treatment (Table 2) [41,50]. In comparison with conventional drugs for the treatment of fistulizing CD, ICERs ascended in excess of 400,000 €/QALY [41] while for newly diagnosed luminal CD IFX was dominant [50]. No CEAs of biologics in UC patients without earlier treatment were found (Table 3).

Cost-Effectiveness of Biologics in Patients with Previous Conventional Medical Treatment
The cost-effectiveness of biologics in CD patients with previous conventional medical treatment was investigated in 12 studies (Table 2) [19,21,22,[42][43][44]46,48,49,51,52,56]. For CD, ICERs for the biologics ranged from dominance to 549,335 €/QALY when compared with those of conventional medical treatment [19,21,22,42,43,48,49]. ADA as an intervention treatment resulted in more frequently lower ICERs than did IFX in comparison with conventional medical treatment [21,22,42,43]. IFX in comparison with surgery was not found to be cost-    [46]. Between biologics cost-effectiveness was investigated in four studies [22,42,52,56]. ICERs above 300,000 €/QALY were seen when comparing IFX with ADA [22,42], while ADA maintenance treatment appeared to be dominant in comparison with IFX maintenance treatment [56]. Two studies evaluated the cost-effectiveness of biologics for different activity levels of CD resulting in more favorable ICERs for severe CD than for moderate CD [21,49]. The cost-effectiveness of biologics for fistulizing CD was examined in two studies (ICERs above 51,000 €/QALY) [44,48] and for luminal CD in two studies (ICERs above 45,000 €/QALY) [46,48]. Biologic induction treatment resulted in lower ICERs than maintenance treatment [21]. In one study, IFX and corticosteroid combination treatment was shown to be cost-effective in comparison with IFX monotherapy [51]. One study found more favorable ICER when including both direct and indirect costs than only direct costs [49]. The ICERs of the studies using lifetime horizon ranged from 11,725 to 947,769 €/QALY [43,44,46].

Cost-Effectiveness of Biologics in Patients with Previous Surgery
The cost-effectiveness of biologics in CD patients having undergone intestinal resection was investigated in two CUAs (Table 2) [39,45]. IFX in comparison with conventional medical treatment was not cost-effective, producing extremely unfavorable ICERs above 1,400,000 €/QALY. No studies investigated the cost-effectiveness of biologics in UC patients with previous surgery (Table 3).

Cost-Effectiveness of Biologics in Patients with Previous Biologic Treatment
The cost-effectiveness of biologics in CD patients with prior biologic treatment was investigated in two CUAs (Table 2) [40,47]. Neither IFX dose escalation in comparison with secondline ADA nor third-line CTZ in comparison with NTZ was cost-effective (ICERs above 300,000 €/QALY). No studies evaluated the cost-effectiveness of biologics in UC patients with prior TNF inhibitor treatment (Table 3).

Effectiveness Data
In all studies, the source of effectiveness was based on at least one randomized controlled trial (RCT). One study used real life data published by specialized inflammatory bowel disease clinics and compared those findings with data from RCTs [54].
In three studies concerning UC [35,36,53], the utility scores were obtained from an UC patient survey carried out in Cardiff Hospital using the EQ-5D and valued using UK tariffs [60]. Utilities were further classified into health states by a Simple Clinical Colitis Activity Index (SCAI). Two studies [22,55] used utilities from patients using Time Trade-off (TTO) valuation technique [61].

Quality Assessment
The results of the quality assessment are shown in  [21,22,37]. The quality elements most commonly omitted from the economic analyses were information on adjustments for data identification, baseline data, treatment effects, data incorporation, and assessment of uncertainty (S2 Table).

Discussion
Altogether, 25 studies were included in this systematic review. The number of the included studies in this review was higher than in previously published reviews for IBD [18][19][20][21][22].  [22,42,55]. A majority of the included studies used IFX or ADA as an intervention treatment, while NTZ and CTZ were investigated only in few studies, and none of the studies considered golimumab. Because of the variability in data input and heterogeneous study designs, the quantitative synthesis of the studies was not possible. On the basis of the current review and willingness-to-pay threshold of 35,000 €/QALY, biologics in comparison with conventional medical treatment and placebo treatment were found to be cost-effective for severe CD in remission induction, while for maintenance treatment cost-effectiveness remained unclear. Biologics were not cost-effective in comparison with surgery for the treatment of severe CD. In moderate CD, biologics did not seem to be cost-effective. Biologics were found not to be cost-effective among CD patients having undergone intestinal resection. ADA was shown to be a more cost-effective biologic treatment option than IFX. Cost-effectiveness between individual biologics remained unclear, however.
Biologics were cost-effective for the treatment of acute exacerbation of severely active UC when compared with either conventional medical treatment, surgery, or placebo treatment. For moderate UC, biologics were not cost-effective. The cost-effectiveness between different biologics remained unclear in UC.
The literature search found five earlier published systematic reviews of the cost-effectiveness of biologics for IBDs [18][19][20][21][22]. In previous reviews, the conclusions have been contradictory and partially unreliable due to a low amount of included CUAs. Four out of five previous reviews evaluated the cost-effectiveness of biologics for CD [18][19][20][21], while one assessed the cost-effectiveness of biologics in both CD and UC [22]. IFX was the only biologic treatment in four reviews [19][20][21][22]. Meanwhile, the latest systematic review by Tang et al included IFX, ADA, CTZ, and NTZ and came to a conclusion that the biologics are cost-effective for CD in certain clinical situations which was congruent with this review [18]. The earlier review by Assasi et al [22] included one CUA [53] showing that scheduled maintenance treatment with IFX is a cost-effective option for UC [22]. The studies included in our review focused mainly The resources consumed. c Productivity costs for the patient and family members. An important issue affecting the conclusions of CEAs relates to the established willingnessto-pay threshold. In the UK, the National Institute of Health and Clinical Excellence (NICE) supports treatments with ICER no higher than 30,000 £ (*35,000 €) per QALY [62], which we used as a threshold in this study. However, there has been much debate as to whether this threshold is too low, and many health care systems have not set a cost-effectiveness threshold at all [54]. The willingness-to-pay threshold commonly used by the Canadian Drug Expert Committee is 80,000 CDN$ (*75,000 €) per QALY [63], while the threshold of 50,000-100,000 US$ (*38,00-75,000 €) per QALY is often used in the US [64,65]. According to the World Health Organization, an intervention is cost-effective if the cost of intervention per QALY is less than three times the country's annual gross domestic product [66]. Even if those thresholds had been used in this review, biologics would not have been deemed cost-effective in most studies. It should be noted that the selection of the willingness-to-pay threshold Most studies used the perspective of the local public health care service or the insurance system while only paying attention to direct costs. Only one study included both direct and indirect costs and reported more favorable ICER when considering both direct and indirect costs in comparison with only direct costs [49]. No clear guidelines exist on how productivity losses should be determined causing concern for the validity of the cost estimates. Included cost components and their valuing methods can be difficult to identify based on the publications. Furthermore, productivity costs included in CUA may cause a risk of double-counting as the impact of morbidity is already included in the calculation of QALY [26]. The patient´s earnings and leisure activities affect variability on the value of the individual´s time [24]. The differences in overall labor costs, health policy, and other health system factors make challenging to compare results between countries. IBDs as chronic diseases are usually diagnosed in early adulthood causing a severe impact on productivity costs. Even though biologics increase the drug costs, they are assumed to improve the health status and to reduce the burden on resources outside the health care system such as absenteeism from work [67]. Consequently, it is appropriate to include indirect costs in CUA, but indirect costs should be presented separately from direct costs [24,26].
When evaluating effectiveness, it is scientifically and ethically important to use the most appropriate alternative treatment as the control group. The comparator with a good efficacy and safety profile should act as the most cost-effective alternative treatment and is usually the intervention most used in clinical practice [32,68]. However, recommendations on the appropriate comparator vary across countries and depend on the research question [68]. A majority of the included studies used the "standard care" or "usual care" as the comparator.
Source of effectiveness data has substantial influence on model results. RCT data was used as effectiveness data in all included CUAs. RCTs give information about efficacy determined in ideal circumstances and cause a risk of overestimating effectiveness in comparison with the treatment in routine healthcare. Therefore, it is plausible to assume that the CUAs using RCTs as a source of effectiveness produce lower ICERs than real-world data. Contrary to that assumption, only one of the included studies derived information from real life studies and resulted in more favorable ICERs when using response rates from real life data rather than from RCTs [54]. However, the uncertainty in economic evaluations, especially in modeling studies, can arise from numerous methodological disagreements among analyses. Uncertainty caused by e.g., using multiple data sources and extrapolation beyond the time horizon of the study involving the use of assumptions was tested by sensitivity analysis in most studies.
In most studies, the source of utility data was reported inadequately and considerable variation existed in the instruments used to collect it. Direct elicitation methods (e.g., SG, TTO) were used more frequently than indirect methods (e.g., EQ-5D). With direct methods patients directly score their preferences for health states and make judgments based on their own relative values, while indirect methods are based on the patients' responses to surveys about various aspects of health states [25]. The methods of direct elicitation can be complex and time consuming. In most cases, indirect utility estimates were obtained by determining the relationship between values on a disease-specific measure to a generic quality of life measure. This is necessary because of the fact that the generic measures have been applied in few studies, while disease specific measures such as CDAI are commonly used in RCTs. The application of different algorithms for conversions creates a further source of heterogeneity in ICER estimates. Because of the variation in the methods used and in the preferences across individuals, the QALYs may vary widely between the studies and this affects the results of the CUAs.
Based on previous literature, studies with longer time horizon produce more favorable ICERs than studies with shorter time horizons [49,53]. As biologics improve patients´health status [67], they have potential to yield gain in terms of reductions in hospitalization, surgeries, and incapacity in future. However, the correlation between the length of the time horizon and cost-effectiveness analyses remains unclear in our study. Although the lifetime horizon is appropriate to capture all health effects and costs for chronic diseases, in most modelling studies the time horizon was limited to one year by the availability of the relevant data and to avoid the bias caused by extrapolation to a longer time horizon.
When considering the previously published systematic reviews, only one study used the standardized quality assessment checklist to evaluate the quality of the included CUAs. As far as we know, this is the first systematic review assessing the quality of economic evaluations by three different checklists. Drummond's checklist is recommended to inform appraisal of the methodological quality of full economic evaluations [32,69]. Drummond's checklist is relevant but not sufficient for modeling studies. Therefore, the modeling's were also assessed using Philip's checklist [32,34,69]. The CHEERS guideline includes additional items relating to the author's disclosure of funding sources and conflicts of interest, sufficient information in article titles, and structured abstracts [33]. The CHEERS guideline evaluates the reporting of the study while Drummond's checklist and Philips' checklist are designed to assess the methodological quality of economic evaluations.
The amount of the fulfilled items according to Drummond's checklist and the CHEERS guideline was higher than using Philips' checklist. The reasons may be aims of the checklist and the extensiveness of Philips' checklist including several topics relevant to modeling studies and not considered in Drummond's checklist and the CHEERS guideline. On average, the same CUAs fulfilled the highest amount of the applicable items according to all three checklists. Most of the studies, which fulfilled most criteria of quality assessment checklists, were HTA reports. Almost half of the included CUAs were funded by the pharmaceutical company or authors had received funding from the pharmaceutical companies during the research project [35,36,42,45,[47][48][49]53,55,56]. Many of the studies funded by the pharmaceutical company produced favorable ICERs [35,36,48,49,53,56]. However, it remained unclear whether the source of funding had an effect on the study results. In addition, the relation between the studies funded by a pharmaceutical company and fulfillment of applicable quality assessment criteria was found to be unclear.
The current review was carefully designed beforehand and documented transparently, improving the validity of the study. The study selection, the data extraction, and the quality assessment were performed by one assessor and any ambiguity was resolved with a second assessor to avoid human mistakes and to improve the reliability of the study. The comprehensive literature search was utilized to minimize bias. The intervention treatments included in the search strategy were limited to biologics that had been granted a marketing authorization by the EMA or FDA for the treatment of IBD. Vedolizumab was not included in the search strategy because its marketing authorization was not granted until the planning and realization of the search strategy was completed.
However, because of a limited amount of available CEAs and some inconsistent results, conclusions remain partially uncertain. Furthermore, variability in data input and heterogeneity in study designs made it challenging to compare studies reliably. To improve the reporting of an individual CEA, it is appropriate to use quality assessment checklists. When using checklists, economic evaluations become more consistent, transparent, and informative. The most important predictors of good cost-effectiveness of the biologics were disease activity, the duration of the biological treatment, and the treatment strategy. Further research is needed to confirm cost-effectiveness in moderate IBD. Future studies should evaluate the cost-effectiveness of all available biologic treatments for IBDs. In addition, CEAs between two different biologics are required to find the most cost-effective treatment strategy for IBD patients.
In conclusion, biologics seem to be cost-effective for the induction treatment of active and severe IBD, but not for the maintenance treatment. Whether there are differences in the costeffectiveness between biologics remains unclear.