Endoscopic ultrasound staging in patients with gastro-oesophageal cancers: a systematic review of economic evidence

Background The sensitivity of endoscopic ultrasound (EUS) in staging gastro-oesophageal cancers (GOCs) has been widely studied. However, the economic evidence of EUS staging in the management of patients with GOCs is scarce. This review aimed to examine all economic evidence (not limited to randomised controlled trials) of the use of EUS staging in the management of GOCs patients, and to offer a review of economic evidence on the costs, benefits (in terms of GOCs patients’ health-related quality of life), and economic implications of the use of EUS in staging GOCs patients. Methods The protocol was registered prospectively with PROSPERO (CRD42016043700; http://www.crd.york.ac.uk/PROSPERO/display_record.php?ID=CRD42016043700). MEDLINE (ovid), EMBASE (ovid), The Cochrane Collaboration Register and Library (including the British National Health Service Economic Evaluation Database), CINAHL (EBSCOhost) and Web of Science (Core Collection) as well as reference lists were systematically searched for studies conducted between 1996 and 2018 (search update 28/04/2018). Two authors independently screened the identified articles, assessed study quality, and extracted data. Study characteristics of the included articles, including incremental cost-effectiveness ratios, when available, were summarised narratively. Results Of the 197 articles retrieved, six studies met the inclusion criteria: three economic studies and three economic modelling studies. Of the three economic studies, one was a cost-effectiveness analysis and two were cost analyses. Of the three economic modelling studies, one was a cost-effectiveness analysis and two were cost-minimisation analyses. Both of the cost-effectiveness analyses reported that use of EUS as an additional staging technique provided, on average, more QALYs (0.0019–0.1969 more QALYs) and saved costs (by £1969–£3364 per patient, 2017 price year) compared to staging strategy without EUS. Of the six studies, only one included GOCs participants and the other five included oesophageal cancer participants. Conclusions The data available suggest use of EUS as a complementary staging technique to other staging techniques for GOCs appears to be cost saving and offers greater QALYs. Nevertheless, future studies are necessary because the economic evidence around this EUS staging intervention for GOCs is far from robust. More health economic research and good quality data are needed to judge the economic benefits of EUS staging for GOCs. PROSPERO Registration Number CRD42016043700. Electronic supplementary material The online version of this article (10.1186/s12885-019-6116-0) contains supplementary material, which is available to authorized users.


Background
Gastro-oesophageal (oesophageal or gastric, or both) cancers (GOCs) are one of the most common cancers in the UK with approximately 16,000 people diagnosed in 2015 [1,2]. Oesophageal and gastric cancers were the seventh and fourteenth most common cause of cancer death respectively in the UK in 2016, as shown from the latest available statistics reported by the Cancer Research United Kingdom (CRUK) [1,2]. It is estimated that a total of around 12,500 people died from these cancers in 2016that is 34 deaths per day [1,2]. Accurate staging of GOCs is vital for determining prognosis and planning appropriate treatment. Accurate staging in the management of GOCs will not only help avoid unnecessary surgical interventions but also will ultimately help reduce the financial pressure on the NHS, which is particularly important given the limited resources available to cancer services and the growing incidence of GOCs [3].
Accurate staging of GOCs can be achieved by a combination of investigative techniques. The techniques used for staging GOC include computer tomography (CT), endoscopic ultrasound (EUS), positron emission tomography (PET) and adjuncts to staging include magnetic resonance imaging (MRI), bronchoscopy, laparoscopy and trans-abdominal ultrasound [4]. CT has been recommended for use at initial staging assessment to determine whether the cancer cells have spread from the primary site of its origin into new areas of the body (i.e. metastasis); but in the absence of metastatic disease, EUS has been advocated as the preferred technique for the assessment and prediction of operability [4]. This is due to the fact that EUS is superior to CT for local regional staging of oesophageal and gastric tumours [4].
Studies and guidelines for the management of oesophageal and gastric cancer have reported that EUS has superior tumour invasion (T) and loco-regional nodal (N) staging ability over CT and PET given its sensitivity, particularly for detection of regional lymph node metastases, although the complementary nature of these investigative techniques must be recognised [5][6][7][8][9][10]. The sensitivity of EUS for staging of GOC has been widely evaluated; however, the economic evidence of EUS staging in the management of GOC patients is scarce. Furthermore, the effectiveness and cost-effectiveness of EUS staging of GOC had not been assessed, particularly in the form of randomised controlled trials (RCT), until the establishment of "COGNATE" trial -a HTA-funded RCT UK study [11].
Given that the economic evidence of EUS for staging of GOC is scant, conducting a systematic review of the economic evidence on EUS staging in patients with GOC is therefore important. It not only gives a meaningful evidencebased insight, from an economic perspective, for researchers and clinical experts in this field but also healthcare commissioners. In view of that, this systematic review aimed to examine all economic evidence (not just from RCTs) of the use of EUS staging in the management of patients with GOC. Systematic reviews of economic evaluations review studies that evaluated both the effectiveness in terms of health effects (usually measured as life-years gained (LYGs) or quality-adjusted-life-years (QALYs), accounting for the quality-of-life outcomes) and cost of the alternative interventions assessed. Economic evaluation is performed by undertaking either a cost-effectiveness analysis (CEA), cost-utility analysis (CUA), cost-consequences analysis (CCA), costbenefit analysis (CBA) or cost-minimisation analysis (CMA). When clinical outcome expressed in natural units (e.g. LYGs, lives saved, improvement in pain score etc) are used as health effects in an economic analysis, this is often referred to as CEA with its parameter of interest being called incremental cost-effectiveness ratio (ICER). Whereas, when QALY, a common unit, is used as health effect in an economic analysis, then this is often referred to as CUA though CEA is preferred by some authors and the resulting parameter of interest is called incremental cost-utility ratio (ICUR). The ICER/ICUR is then compared with the official or approximate willingness to pay (WTP) ceiling ratio for a unit of effect, that is, threshold used for decision making. CCA reports costs and outcomes in disaggregated form for each alternative [12]. CBA converts clinical outcomes into monetary units so that a net benefit (or cost) can be estimated [12]. CMA measures which alternative has the least cost, this method is only applied when the outcomes of alternative interventions have been proven to be equivalent. The protocol of this systematic review was registered prospectively with PROSPERO, an international prospective register of systematic reviews (Registration number 2016:CRD42016043700; http://www.crd.york.ac.uk/PROSPERO/display_record. php?ID=CRD42016043700) [13]. This paper offers a review of economic evidence on the costs, benefits (in terms of GOC patients' health-related quality of life), and economic implications of the use of EUS for staging GOC patients.

Methods
This review was carried out and reported in accordance with the published updated Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [14,15].

Searches and study selection
Searches for this systematic review were conducted using a range of electronic databases: MEDLINE (ovid), EMBASE (ovid), The Cochrane Collaboration Register and Library (including Cochrane Central Register of Controlled Trials (CCRCT), Cochrane Reviews, Database of Abstracts of Reviews of Effects (DARE), Health Technology Assessment (HTA), British National Health Service Economic Evaluation Database (NHS EED), Cochrane Methodology Register (CMR)), CINAHL (EBSCOhost), Web of Science (Core Collection). Searches were restricted to publications from the last 20 years (1996-2016) as per the registered protocol on PROSPERO (Registration number 2016:CRD42016043700) [13]. To ensure that the review was as up-to-date as possible, the searches were re-run on all databases to cover 2016-2018 (search update on 28/04/2018).
In order to ensure a comprehensive search was achieved and any relevant research had not been missed, online searches were also conducted through the following internet search engines and appropriate websites to identify grey literature, reports, ongoing and unpublished studies from conference papers and abstracts: The reference lists of papers that were included in the review were searched for further publications that had not been identified in the electronic searches. Contacts with study authors were made to locate further relevant literature and publications.
Guided by the review objectives, the search terms as shown in Table 1 were developed using the PICO framework [16,17]. The PICO framework was utilised to help shape, design and construct the search process to identify all relevant published and unpublished materials from various sources. Titles, abstracts and full-text papers were searched for using these search terms.
The search strategy for each of the five electronic databases was developed, checked and tested by an information specialist before finalising the search terms; this process was informed by the search strategy of a wider evidence synthesis that includes a systematic review of non-economic studies of treatments for resectable adenocarcinoma of the stomach, gastro-oesophageal junction and lower oesophagus [18]. An example of search strategy used in the Medline Ovid database is as shown in the additional file (see Additional file 1). Table 2 presents the inclusion and exclusion criteria, using the economic evidence review design framework outlined in the University of York Centre for Reviews and Dissemination (2009) [12]: Population, Interventions, Comparators, Outcomes, and Type of Evidence. Due to resources constraints, only studies written in English were included. This includes international studies that have been translated or written in English.

Data extraction
Titles and abstracts of all studies identified were screened and assessed for relevance against the inclusion criteria by two independent reviewers (STY and NB). The inclusion or exclusion of each study was checked and confirmed. All potentially relevant full-text papers were then obtained and screened against the inclusion criteria, with disagreements resolved through discussion until agreements were achieved collectively. Disagreements occurred when for example the reviewers had different views on whether a retrieved paper should be included in the review.
Following screening, relevant information from all fulltext papers included in the review were extracted by the primary reviewer (STY) using an adapted standardised form [12], and checked by the second reviewer (NB). Two adapted standardised forms were developed and used for data extractionone for economic studies and another for economic modelling studies.

Quality assessment
The quality of all full-text papers included in the review were assessed and rated independently by the two reviewers using the Critical Appraisal Skills Programme (CASP) economic evaluation checklist [19] tool for economic studies and the Philips et al's economic modelling checklist [20] tool for economic modelling studies. The papers were critically appraised to assess to what extent the content of these papers complied with the criteria of good practice in economic evaluation and if there was any obvious bias. Disagreements between the reviewers were resolved through discussion until agreements were achieved collectively. Disagreements occurred when for example the reviewers had different score on an included paper.

Data synthesis
All studies included in the review were summarised and compared across studies in a narrative form to answer the review objectives. The aims, methods, and results of the studies reviewed were synthesised narratively. This demonstrates the heterogeneity of the studies in terms of characteristics [12]. Due to the heterogeneity of the studies in terms of the study type and outcomes across the studies, meta-analysis was not appropriate [12]. Costs were converted into British pounds sterling, £, using the appropriate exchange rate published in the International Monetary Fund [21] and inflated to 2017 price year using the hospital and community health services (HCHS) index [22][23][24][25] for the studies included in the review.

Literature search: identification of studies
Overall, the search from 1996 to 2016 identified 197 potentially relevant studies, six of which fulfilled the  inclusion criteria and were included in the review ( Fig. 1). Of the six studies included, three were economic analysis studies and three were economic modelling studies.
To ensure that the review was as up-to-date as possible, the searches were re-run on all databases to cover 2016-2018 (search update on 28/04/2018); 30 potentially relevant papers were identified but none met the inclusion criteria. In such case, the final number of studies included in the review remained at six.

Study descriptions
Tables 3 and 4 summarises the characteristics of the six studies included in the review. There were three economic analysis studies (Table 3) and three economic modelling studies ( Table 4). Five of the studies included in the review were US studies, and one was a UK study. Of the three economic analysis studies, two were cost analyses [26,27] and one was a cost-effectiveness analysis [11]. Of the three economic modelling studies, two were cost-minimisation analyses [29,30] and one was a cost-effectiveness analysis [31]. All of the three economic modelling studies used decision tree modelling techniques to explore staging strategies.
The six studies included in the review differed quite markedly in terms of their design. Only one study used primary cost and outcome data collected in prospective evaluation [11], one study used data collected in prospective case series [27], one study used retrospective data [26], and the remaining three studies synthesised data from secondary sources in a decision tree model [29][30][31]. Of the six studies, only one [11] was a randomised controlled trial and included participants diagnosed with gastro-oesophageal cancer (i.e. oesophageal, gastro-oesophageal junction or gastric cancer); the other five were non-trial studies and included participants diagnosed with oesophageal cancer. Amongst the six studies, Russell et al. (2013) [11] was again the only study which evaluated costs of health care resource use covering secondary care contacts and hospital prescribed drugs in addition to cost of EUS, collected prospectively in the trial. In terms of health outcome measures, two studies [11,31] included quality-adjusted life year (QALY) as the measure of effect and conducted a cost-effectiveness analysis to assess the gain in QALYs relative to the costs of different staging strategies. The remaining four studies [26,27,29,30] did not explore QALY or other quality of life measures but only cost.

Quality assessment
Each of the six studies included in the review were critically appraised against the appropriate source of quality appraisal checklist: the CASP economic evaluation checklist [19] was used for the three economic studies, and Philips et al's economic modelling checklist [20] was used for the remaining three economic modelling studies. Table 5 and Table 6 summarised the quality assessment of the three economic studies and three economic modelling studies, respectively. Table 5 shows the study quality of the three economic studies was generally good, scoring on average greater than 75%, although only one study [11] met all quality criteria on the CASP economic evaluation checklist. The study by Shumaker et al. (2002) [26] scored the second highest, followed by Chang et al. (2003) [27]. Of these three economic studies, two had missing key information: Chang et al. (2003) [27] reported neither cost perspective, cost inflation, discounting nor price year, and sensitivity analysis was not undertaken; likewise, Shumaker et al. (2002) [26] did not state whether their reported costs were discounted or inflated as appropriate. Table 6 shows the study quality of the three economic modelling studies included in the review was  [29]. One study [29] did not state the perspective of the model and all three [29][30][31] did not specify the time horizon of the decision tree model. There was insufficient detail of how parameters in the model were identified [31] and how data were modelled [30]. There was also a lack of clarity with regards to the source of probabilities and cost data used in the decision tree model [29].

Data synthesis results
All of the six studies included in the review exhibit EUS as a complementary imaging technique to other imaging modalities such as CT and PET scanning for staging gastro-oesophageal cancer. This is in agreement with a previously published meta-analysis study of diagnostic test characteristics for EUS, CT, and PET scanning [8], concluding that the three approaches were complementary. Results from three of the economic studies [11,26,27] show staging of oesophageal or gastro-oesophageal cancer with EUS could potentially save costs. Similarly, results from two of the modelling studies [29,30] show that EUS or EUS-fine-needle aspiration biopsy (FNA) is the least costly staging technique for oesophageal cancer. The study by Wallace et al. (2002) [31] shows that EUS-FNA in addition to CT scan is the least costly strategy than all other strategies i.e. CT alone, CT+ thoracoscopy and laparoscopy (TL), CT + EUS-FNA + TL, CT + PET+EUS-FNA and PET+EUS-FNA.
Results from the two studies [11,31] in which qualityadjusted life year (QALY) and cost data were available demonstrate the use of EUS [11] or EUS-FNA [31] as an additional staging technique for gastro-oesophageal cancer offered more QALYs and costed less, on average, compared to staging techniques without EUS. Russell et       [11] reported that EUS resulted in a QALY gain of 0.1969 QALYs and saved costs by £2860, on average, per patient (£3364 per patient, 2017 price year); combining these benefits and savings demonstrates that EUS is likely to be cost-effective with a probability of 96% at the UK NICE's threshold of £20,000-£30,000 per QALY [28]. Similarly, Wallace et al. (2002)'s [31] modelling study showed that using EUS-FNA as an additional staging technique offered greater QALYs and saved more costs, on average, than staging strategy without EUS.  (£66,588 per QALY, 2017 price year), a ratio that is less than that of other medical treatments but above accepted thresholds in the USA and UK.

Main findings
This systematic review of economic evidence of EUS staging in patients with GOC revealed a considerably small number of relevant studies. Studies varied in quality, study design and method. Study quality was generally satisfactory across  all the studies included in the review, but only one of these studies [11] met all reporting and quality criteria. Differences in study design make it difficult to draw definitive conclusions as to whether the use of EUS as an additional staging technique could be considered cost-effective i.e. value for money which can be assessed by comparing the costs (monetary term) and health effects (non-monetary term) of an intervention with the alternative. Health effect of an intervention is usually measured in terms of QALYs (Quality-Adjusted Life Years), a summary measure of health outcome and also a common unit used for economic evaluation of an intervention, as recommended by the UK's NICE (The National Institute for Health and Care Excellence) [28]. Given the differences in study design, a head-to-head comparison of the results couldn't be made from the Russell [31] studies to draw definitive conclusions. Although both of these studies had evaluated both costs and QALYs, their respective study designs were too different to allow direct comparison; one was an economic evaluation study using primary data [11] and the other an economic modelling study using secondary data [31]. Nevertheless, the economic evidence identified in this review, especially the better quality studies, provided useful findings on the value of EUS staging in the management of GOC patients, which could be of importance to policymakers and healthcare commissioners. Among the six studies included in the review, two studies [11,31] are the most robust in terms of including and comparing the relative costs and QALYs of different staging strategies, for example GOC staging with and without EUS. Findings from both of these two studies demonstrated that use of EUS as an additional imaging technique could save costs and offer greater QALY gains. This could be due to the fact that EUS has been known to be beneficial in terms of its sensitivity for locoregional staging of GOC [4,6,8,[32][33][34]. For that reason, using EUS as a complementary imaging technique to other imaging techniques such as CT and PET scanning for staging GOC could undoubtedly help minimise unnecessary treatments [4,35,36]; and thus potentially could save costs and offer greater health benefits to patients in terms of QALY gains. The EUS cost saving evidence was also supported by the remaining four studies [26,27,29,30] evaluating only the cost of EUS e.g. whether EUS is a cost saving strategy or the least costly staging strategy. Russell et al. (2013) [11] further argued that EUS has a considerably high probability of being cost-effective under current recommended UK NICE's threshold of £20,000 to £30, 000 per QALY [28]. Thus, despite the scarcity of economic evidence in this field, from these studies identified in the review, there is some positive economic evidence relating to the cost-effectiveness of EUS in the management of patients with GOC.

Strength and limitations of review methods
This review adds to the literature by providing critical evaluation of the health economics evidence of EUS staging in gastro-oesophageal cancers (GOCs), for which there is a lack of well-conducted economic studies. Though a systematic review in this field was published 20 years ago [37], this systematic review is the most upto-date collection of economic literature in this area. Twenty years on since the review by Harris et al. (1998) [37], still only six papers were found in the area of health economics of EUS staging in GOC. This shows that there is a lack of prioritisation of research in this area.
Broad search terms were used to develop a comprehensive search strategy for each of the databases used in this systematic review. The resultant retrieved studies were quality appraised, using both the published standard checklists recommended for use in assessing the quality of economics articles in systematic reviewthe Critical Appraisal Skills Programme (CASP) economic evaluation checklist [19] and the Philips et al's economic modelling checklist [20] for the retrieved economic studies and economic modelling studies, respectively. The narrative summary of the review not only described the economic evidence of EUS staging in GOC but also served as a platform for providing a holistic insight into the health economics research available to date in this area. The latter is particularly helpful for commissioners, clinicians and researchers to elicit information and potentially to facilitate the development of further research in this area. This review has several limitations. Heterogeneity of the included studies in the review in terms of study designs and methods meant that a meta-analysis of studies was not possible and a narrative summary was used. We also acknowledge that different countries have different thresholds for both the investigation of, and surgical management of gastro-oesophageal malignancies. This can result in variation between practices and hence difficulty in translating financial recommendations across regions. In terms of impact that EUS has on patients' quality of life and its costs, the lack of the availability of health economics research in this area means that it is considerably difficult, particularly for commissioners and clinicians, to guide evidence-based practice from an economic perspective.

Further research
This systematic review shows that the economic evidence available to date in this area is still scarce. There was a lack of health economic research collecting data, especially primary data, on both costs and effects (such as utility values to construct QALYs) of EUS staging in GOC. To improve this, there is a need for more primary health economic research in this area, particularly integrated clinical and economic trials of EUS staging in GOC that can offer robust evidence of costs and effects.

Conclusions
Despite the lack of economic evidence on costs and benefits of EUS staging for GOC, the data available from this review suggest use of EUS as a complementary staging technique to other staging techniques for GOC appears to be cost saving and offers greater QALYs. Based on the only randomised controlled trial conducted in the UK identified in this review, EUS seems to have high probability of being cost-effective at the UK NICE's threshold of £20,000-£30,000 per QALY. Nevertheless, future studies are necessary because the economic evidence around EUS staging interventions for GOC is far from robust. More health economic research and good quality data are needed to judge the economic benefits of EUS staging for GOC, particularly primary health economic research that collects primary data on the costs and effects (such as QALYs) of EUS staging in GOC.