Research on the cost-effectiveness of upper digestive endoscopy for the diagnostic of early gastric cancer

2014 2 3 Dissertação de candidatura ao grau de Doutor apresentada à Faculdade de Medicina da Universidade do Porto Título: Estudo para a Investigação do Custo – Eficácia da Endoscopia Digestiva Alta no Diagnóstico Precoce de Cancro Gástrico Title: Research on the cost-effectiveness of upper digestive endoscopy for the diagnosis of early gastric cancer Supervisors:  Mário Dinis-Ribeiro, MD, PhD (University of Porto) CINTESIS - Center for Research in Health Technologies and Information Systems, Faculty of Medicine, Porto University Gastroenterology Department, Portuguese Oncology Institute – Porto  Francisco Rocha Gonçalves, PhD (University of Porto) CINTESIS - Center for Research in Health Technologies and Information Systems, Faculty of Medicine, Porto University Portuguese Oncology Institute – Porto Thesis committee:  Mário Jorge Dinis-Ribeiro, PhD, University of Porto  Pedro Manuel Narra Figueiredo, PhD, University of Coimbra  Carla Rolanda da Rocha Gonçalves, PhD, University of Minho  Manuel Guilherme Gonçalves Macedo, PhD, University of Porto  Pedro Filipe Vieira Pimentel-Nunes, PhD, University of Porto  Luís Filipe Ribeiro Azevedo, PhD, University of Porto Art.º 48º, § 3º - “A Faculdade não responde pelas doutrinas expendidas na dissertação” (Regulamento da Faculdade de Medicina da Universidade do Porto, Decreto-Lei nº 19 337 de 29 de Janeiro de 1931) 4 5 Professores Catedráticos Alberto Manuel Barros da Silva Altamiro Manuel Rodrigues Costa Pereira António Albino Coelho Marques Abrantes Teixeira António Carlos Freitas Ribeiro Saraiva Daniel Filipe Lima Moura Deolinda Maria Valente Alves Lima Teixeira Francisco Fernando Rocha Gonçalves Isabel Maria Amorim Pereira Ramos João Francisco Montenegro Andrade Lima Bernardes Joaquim Adelino Correia Ferreira Leite Moreira José Agostinho Marques Lopes José Carlos Neves da Cunha Areias José Eduardo Torres Eckenroth Guimarães José Henrique Dias Pinto de Barros José Manuel Lopes Teixeira Amarante José Manuel Pereira Dias de Castro Lopes Manuel Alberto Coimbra Sobrinho Simões Manuel Jesus Falcão Pestana Vasconcelos Maria Amélia Duarte Ferreira Maria Dulce Cordeiro Madeira Maria Fátima Machado Henriques Carneiro Maria Leonor Martins Soares David Patrício Manuel Vieira Araújo Soares Silva Raquel Ângela Silva Soares Lino Rui Manuel Almeida Mota Cardoso Rui Manuel Lopes Nunes 6 7 Professores Jubilados e Aposentados Abel José Sampaio da Costa Tavares Abel Vitorino Trigo Cabral Alexandre Alberto Guerra Sousa Pinto Álvaro Jerónimo Leal Machado de Aguiar Amândio Gomes Sampaio Tavares António Augusto Lopes Vaz António Carvalho Almeida Coimbra António Fernandes da Fonseca António Fernandes Oliveira Barbosa Ribeiro Braga António Germano Pina Silva Leal António José Pacheco Palha António Manuel Sampaio de Araújo Teixeira Belmiro dos Santos Patrício Cândido Alves Hipólito Reis Carlos Rodrigo Magalhães Ramalhão Cassiano Pena de Abreu e Lima Daniel Santos Pinto Serrão Eduardo Jorge Cunha Rodrigues Pereira Fernando Tavarela Veloso Francisco de Sousa Lé Henrique José Ferreira Gonçalves Lecour de Menezes Jorge Manuel Mergulhão Castro Tavares José Carvalho de Oliveira José Fernando Barros Castro Correia José Luís Medina Vieira José Manuel Costa Mesquita Guimarães 8 Levi Eugénio Ribeiro Guerra Luís Alberto Martins Gomes de Almeida Manuel António Caldeira Pais Clemente Manuel Augusto Cardoso de Oliveira Manuel Machado Rodrigues Gomes Manuel Maria Paula Barbosa Maria da Conceição Fernandes Marques Magalhães Maria Isabel Amorim de Azevedo Mário José Cerqueira Gomes Braga Serafim Correia Pinto Guimarães Valdemar Miguel Botelho dos Santos Cardoso Walter Friedrich Alfred Osswald 9 Ao abrigo do Art.º 8º do Decreto-Lei n.º 388/70, fazem parte desta dissertação os seguintes trabalhos publicados ou em publicação: I. Areia M, Carvalho R, Cadime AT, Rocha Gonçalves F, Dinis-Ribeiro M. Screening for Gastric Cancer and Surveillance of Premalignant Lesions: a Systematic Review of Cost-Effectiveness Studies. Helicobacter 2013; 18: 325-37. II. Marques-Silva L, Areia M, Elvas L, Dinis-Ribeiro M. Prevalence of Gastric Precancerous Conditions: A Systematic Review and Meta-analysis. Eur J Gastroenterol Hepatol 2014; 26: 378–87. III. Areia M, Alves S, Brito D, Cadime AT, Carvalho R, Saraiva S, Ferreira S, Moleiro J, Dias Pereira A, Carrasquinho J, Lopes L, Ramada J, Marcos-Pinto R, Pedroto I, Contente L, Eliseu L, Vieira AM, Sampaio M, Sousa HT, Almeida N, Gregório C, Portela F, Sofia C, Braga V, Baginha E, Bana e Costa T, Chagas C, Lebre Mendes L, Magalhães-Costa P, Matos L, Rocha Gonçalves F, Dinis-Ribeiro M. Health-related Quality of Life and Utilities in Gastric Premalignant Conditions and Malignant Lesions: a Multicentre Study in a High Prevalence Country. J Gastrointestin Liver Dis December 2014 Vol. 23 No 4. IV. Areia M, Dinis-Ribeiro M, Rocha Gonçalves F. Cost-utility analysis of endoscopic surveillance of patients with gastric premalignant conditions. Helicobacter 2014 Aug 28. doi: 10.1111/hel.12150. Em cumprimento do disposto no referido Decreto-Lei o aluno declara que participou ativamente em todos os trabalhos acima referidos. 16 17 Preamble The focus of research for the present Thesis was to establish the best cost-utility estimate for the Portuguese population on the use of conventional upper endoscopy for the diagnosis of early gastric cancer. In light of this, our main aim was to perform an economic analysis on the secondary prevention of gastric adenocarcinoma by endoscopic surveillance of patients at high risk of progressing from extensive atrophy or intestinal metaplasia conditions to cancer. The study was designed for the Portuguese population because Portugal has the highest incidence of the disease in Western Europe. The knowledge obtained from this project is very important for a Portuguese Gastroenterologist that works in an Oncology Institute, not only by answering some unsolved questions in the clinical practice pointed out by recent guidelines, such as follow-up or not patients at-risk, how, when and at what cost, but also allowing to use the obtained know-how in other fields of Gastroenterology. The results of the study should be of interest at a national level as the prevalence of gastric premalignant conditions is relevant in our country and most gastroenterologists in Portugal perform upper endoscopic exams in their clinical practice. Furthermore, the results of the model obtained could be replicated in other countries as the clinical model should be applicable in other settings, just needing the proper adjustments on local costs, providing evidence that could be generalized to gastroenterologists worldwide, interested in the problem of gastric cancer. 18 Along with the four specific studies necessary for the present Thesis, the author also promoted the work in this field by dissemination of proposals publishing an opinion article in an indexed Portuguese journal (Areia, Pimentel-Nunes, Marcos-Pinto, & Dinis-Ribeiro, 2013), promoting the search for the best evidence by publishing an article on quality of endoscopic reporting in Gastroenterology (Areia, Soares, & DinisRibeiro, 2010), participating in a gastric cancer book chapter promoted and published by the Portuguese Medical Association (Marcos-Pinto, Areia, Pimentel-Nunes, & DinisRibeiro, 2013), performing a national multicentre cross-sectional study to assess the prevalence of gastric premalignant conditions and general performance of endoscopy in our country (Areia, Dinis-Ribeiro, & Portuguese Society of Digestive Endoscopy, 2014) and participating as a co-author of the guidelines on the Management of Precancerous conditions and lesions in the Stomach (MAPS) (Dinis-Ribeiro, Areia et al., 2012; Dinis-Ribeiro, Areia et al., 2012). 19 Outline of Thesis In Chapter I, the rationale to the subject chosen for the present thesis is presented. A brief introduction on the gastric cancer problem worldwide and specifically in Portugal is offered, along with the available evidence on economic studies on surveillance of premalignant conditions and clinical guidelines recommendations by the time of the thesis origin. In Chapter II, the background that explains the importance of the gastric cancer theme is presented along with a description of the gastric carcinogenesis cascade assumptions, the relevance of endoscopy for the diagnosis, assessment of extension and surveillance of gastric premalignant conditions and how economic studies are implemented in the medical field taking in consideration the published guidelines for this specific type of studies, justifying our choice on a cost-utility model for the endoscopic surveillance of gastric premalignant conditions. In Chapter III, the aims of each of the four studies performed along the present thesis are presented. In Chapter IV, we present the publications that emerged from the present thesis, in its final published format. In Chapter V, a discussion of all the results of our data is presented. Each result is discussed in comparison with the available evidence and the cost-utility model is also compared with previous similar reports in terms of strength of evidence and dissimilarities. 20 Finally, in Chapter VI, our thesis conclusions are presented and further research is identified. 21 Summary 22 23 Gastric adenocarcinoma represents a health problem worldwide due to its high incidence and mortality rates being the fourth most common malignancy and the second leading cause of cancer death (Ferlay, Shin et al., 2010). Its prognosis is highly dependent of the stage at diagnosis, that usually occurs in advanced stages, requiring demanding treatments and costs (Hundahl, Phillips, & Menck, 2000). Although being a worldwide problem, different prevalence rates among countries lead to different options; in Japan, universal screening of the population using fluoroscopy is used, followed by conventional endoscopy with biopsies for the positive results (Hamashima, Shibuya et al., 2008) whilst in most other developed countries the low incidence rate makes screening not cost-effective (Hirota, Zuckerman et al., 2006). In Portugal, the incidence rate for gastric cancer is located in-between these two realities and so, it is not straightforward that in our country such a program would be justifiable. This type of carcinoma develops after years of progression of a benign condition such as non atrophic gastritis that might evolve to premalignant conditions, namely chronic atrophic gastritis and intestinal metaplasia, before the development of dysplasia and invasive cancer (Correa, 1992). Endoscopy is usually the first exam to perform for the diagnosis of these gastric premalignant conditions (Hirota, Zuckerman et al., 2006). Based on the assumption that chronic gastritis starts in the antrum and then spreads to the corpus in an upward manner so that conditions of the corpus are more extensive and related to more advanced stages, it has been suggested to discriminate this extensive phenotype as it could imply an increased risk of progression to dysplasia and invasive cancer. As so, recent guidelines suggest a 3-yearly endoscopic follow-up for these patients, based on clinical data but with no support from the published 24 economic studies due to conflicting results (Dinis-Ribeiro, Areia et al., 2012; DinisRibeiro, Areia et al., 2012). Having Portugal an intermediate-risk population for gastric cancer (GLOBOCAN, 2012), to improve our patients quality of live and overall survival, it is important to improve our rates of early gastric cancer detection in order to achieve more curative treatments by endoscopic resections techniques or detection of invasive cancers in an early stage, treatable with less debilitating treatments. As so, the objective of the present thesis was to determine the cost-utility of endoscopic surveillance every 3 years of patients with extensive gastric premalignant conditions compared with no surveillance. To accomplish that goal we first performed two systematic reviews and one cross-sectional study in order to obtain the best available clinical data for the Portuguese population and then a cost-utility economic analysis on this hypothesis. Using a Markov model to compare the two strategies and using a societal perspective, clinical data was collected from the mentioned systematic reviews of the literature, costs from published national data and community utilities derived from the crosssectional study by using the EuroQol questionnaire in terms of Quality Adjusted Life Years (QALY) (EuroQol, 1990). For the Markov model, the population started at the age of 50, a time horizon was set for 25 years and an annual discount rate of 3% was used for both costs and effectiveness. The results of the model showed that endoscopic surveillance every 3 years provided an Incremental Cost-Effectiveness Ratio (ICER) of € 18,336 below the adopted threshold of € 36,575 proposed by economic guidelines and this strategy dominated 25 surveillance every 5 or 10 years. Even when the model was evaluated in sensitivity analysis, only a few utilities proved to be relevant in deterministic analysis, while probabilistic analysis showed that in 78% of simulations the model remained costeffective. Thus, the conclusion of this thesis is that endoscopic surveillance of patients with premalignant conditions such as extensive atrophy or intestinal metaplasia, every 3 years, in an intermediate-risk country such as Portugal, is cost-effective. 32 33 Chapter I - Rationale 34 35 Gastric cancer is a health problem presenting very different degrees of magnitude worldwide. Its' geographical incidence varies considerably, being particularly marked in Eastern Asia, Central and Eastern Europe, and South and Central America and although a decrease in its incidence is being observed, its burden is still unacceptably high (Ferlay, Shin et al., 2010). In Portugal it ranks fifth in cancer deaths and its incidence is the highest in Western Europe (Ferlay, Parkin, & Steliarova-Foucher, 2010). When cancer is diagnosed so late, patients are offered treatments like surgery, chemotherapy or radiotherapy that are costly, impair quality of life, and still offer a poor prognosis, with an overall 5 year survival rate below 25% (Ferlay, Shin et al., 2010; Hundahl, Phillips, & Menck, 2000; Karim-Kos, de Vries et al., 2008). Specifically in Portugal, gastric cancer is a neoplasm with a high incidence rate and most cases also present in advanced stages conducting to poor prognosis. To prevent late detection of the disease, health authorities should choose between primary prevention programs such as Helicobacter pylori (H. pylori) screening and eradication to prevent the biologic evolution from normal gastric mucosa to premalignant conditions and to invasive cancer or secondary prevention programmes like population screening or surveillance of patients at high risk for the development of the disease to allow gastric cancer detection in earlier stages with more favourable survival, preferably supported by economic analysis for the strategy adopted (Areia, Carvalho, Cadime, Rocha Goncalves, & Dinis-Ribeiro, 2013; Dinis-Ribeiro, Areia et al., 2012; Dinis-Ribeiro, Areia et al., 2012; Malfertheiner, Megraud et al., 2012). Secondary prevention is usually based on endoscopy because it is widely available, accurate, involves relatively minor invasiveness and offers the chance of simultaneously 36 performing diagnostic biopsies and/or therapeutic procedures (Hirota, Zuckerman et al., 2006). This choice of a screening or surveillance strategy depends on the incidence of the disease and while in Japan endoscopic screening is considered cost-effective, in most countries the balance between endoscopic costs and gastric cancer treatment gains renders screening not cost-effective (Areia, Pimentel-Nunes, Marcos-Pinto, & DinisRibeiro, 2013; Dinis-Ribeiro, Areia et al., 2012; Dinis-Ribeiro, Areia et al., 2012; Hamashima, Shibuya et al., 2008). The intestinal subtype of gastric adenocarcinoma is known to be preceded by a cascade of premalignant conditions and lesions, namely chronic atrophic gastritis, intestinal metaplasia and gastric dysplasia (Correa, 1988; Lauren, 1965) and several different classifications exist for their histological classification (Capelle, de Vries et al., 2010; Dixon, Genta, Yardley, & Correa, 1996; Rugge, Correa et al., 2002; Rugge, Meggio et al., 2007). One reason for this late detection in advanced stages of disease might be the fact that, until 2012, there were no international recommendations to guide clinicians in their care of individuals with these changes and this lead to a wide heterogeneity of practice and failure to diagnose patients with curable forms of cancer in most countries (de Vries, van Grieken et al., 2008; Dinis-Ribeiro, Lopes, da CostaPereira, & Moreira-Dias, 2008). Based on the assumption that chronic gastritis starts in the antrum and then spreads to the corpus in an upward manner so that conditions of the corpus are more extensive and related to more advanced stages, it has been suggested to discriminate this extensive phenotype at an increased risk of progression to dysplasia and invasive 37 cancer (Dinis-Ribeiro, Areia et al., 2012). The surveillance of patients with these lesions might allow the endoscopic treatment of dysplastic lesions or the detection of early invasive cancers that are treatable with less demanding and cheaper treatments with better prognosis (Correa, 1992; Dinis-Ribeiro, Lopes et al., 2004; Dinis-Ribeiro, Pimentel-Nunes et al., 2009; NCCN, 2013). Recently published guidelines suggested a 3-yearly endoscopic surveillance for the follow-up of patients with high risk conditions of progression, such as the presence of extensive atrophy or intestinal metaplasia (Dinis-Ribeiro, Areia et al., 2012; DinisRibeiro, Areia et al., 2012) but only three studies have been published so far on the subject. These three studies published on the surveillance of patients with extensive conditions reported conflicting results on the cost-effectiveness of this option: one concluding on the cost-effectiveness of annual surveillance of these patients but the other two providing incremental cost-effectiveness ratios above the accepted threshold, probably related to different assumptions on rates of progression of these conditions (Dinis-Ribeiro, da Costa-Pereira, Lopes, & Moreira-Dias, 2007; Hassan, Zullo et al., 2010; Yeh, Hur, Kuntz, Ezzati, & Goldie, 2010). One of these studies represents the only cost-effectiveness study ever conducted on the gastric cancer issue in Portugal, and consisted of a specific endoscopic technology (magnification chromoendoscopy) on a yearly basis for patients at-risk for gastric cancer (patients with chronic atrophic gastritis and intestinal metaplasia) (DinisRibeiro, da Costa-Pereira, Lopes, & Moreira-Dias, 2007). For the use of conventional upper endoscopy on a 3-yearly surveillance protocol for high-risk patients as suggested 38 from the guidelines, no such study was ever conducted raising the need for the present research. 39 Chapter II - Background 40 41 Gastric cancer burden Gastric cancer is a worldwide problem due to its high mortality as in most cases patients present in advanced stages of the disease. It is the fourth most common malignancy and the second leading cause of cancer death just after lung cancer representing 9.7% of the total mortality with a estimated number of 990,000 new cases per year and 738,000 deaths (Ferlay, Shin et al., 2010) and although a decrease in its incidence is being observed (it ranked first worldwide in 1975), its burden is still unacceptably high (Ferlay, Shin et al., 2010; Parkin, Stjernsward, & Muir, 1984). In Portugal it ranks fifth in cancer deaths and its incidence is the highest in Western Europe with 3,018 cancers diagnosed in 2012 (Ferlay, Parkin, & Steliarova-Foucher, 2010; GLOBOCAN, 2012). Most cases are related to infection with Helicobacter pylori and other environmental agents and so incidence increases with age. Due to the expected increase aging of populations, namely in Portugal, the estimates for the next 20 years will have an increase in incidence and mortality compared to the present time, empathising the burden of this disease for the next decades (GLOBOCAN, 2012) (Figure 1). 48 Other factors are related to progression of these conditions to invasive cancer but their role in the clinical practice is still questionable and difficult to apply in the clinical practice decisions such as the strain of the H. pylori, duration of infection, presence or absence of other environmental risk factors (e.g., poor diet, smoking, salt), host genetic factors, in particular genes that regulate expression of pro inflammatory agents (interleukin 1 β, tumour-necrosis factor) and some innate immunity receptors such as toll-like receptors (TLRs) (Abrams & Wang, 2010; Kuipers, Perez-Perez, Meuwissen, & Blaser, 1995; Machado, Figueiredo et al., 2003; Machado, Pharoah et al., 2001; Pereira, Sousa et al., 2006; Pimentel-Nunes, Afonso et al., 2011; PimentelNunes, Goncalves et al., 2013; Pinto-Correia, Sousa et al., 2006). Familial aggregation of gastric cancer also might have some role in about 10% of cases suggesting that a specific surveillance might be desirable, although this is still not consensual (MarcosPinto, Carneiro et al., 2012; Marcos-Pinto, Dinis-Ribeiro et al., 2012). 49 Endoscopy Upper digestive endoscopy is considered the ideal procedure for the diagnosis of upper digestive system diseases involving the oesophagus, stomach and duodenum. Its widespread availability, the improved accuracy for most diseases, its relatively minor invasiveness and the possibility of simultaneously performing diagnostic biopsies and/or therapeutic procedures usually makes it the first choice for the study of upper digestive diseases (Hirota, Zuckerman et al., 2006). It is usually the first exam performed for the diagnosis of gastric premalignant conditions in patients with gastric complaints but its cost precludes its use as screening tool for gastric cancer in most clinical scenarios. Even though some efforts at developing screening or diagnostic methodologies to determine the presence and extent of atrophy or intestinal metaplasia have been made, endoscopy alone is said to be inaccurate to determine these changes and the gold standard for diagnosis is histology from biopsies taken during upper endoscopy (Areia, Amaro et al., 2008; Areia, Amaro et al., 2008; Dinis-Ribeiro, da Costa-Pereira et al., 2003). As Portugal is the Western European country with the highest incidence of gastric cancer it is crucial to have data on prevalence of gastric premalignant conditions (Ferlay, Shin et al., 2010; Hundahl, Phillips, & Menck, 2000). Furthermore, patient acceptance to undergo an endoscopy and the manner in which these exams are performed in terms of associated techniques, use of sedation and complications are mandatory to quantify costs that might be relevant in further economic studies that 50 consider endoscopy for population screening or follow-up of asymptomatic at-risk patients in Portugal. Economic studies Economic studies in medicine are usually designed to compare technologies or clinical strategies by simultaneously addressing their differences in terms of both clinical benefits and the cost involved in achieving those gains (Drummond, Sculper, Torrance, O'Brien, & Stoddart, 2005). The objective is to compare different technologies to achieve the same purpose based on the assumption that financial resources are limited and options have to be made incorporating not only clinical effectiveness but also costs. This concern is mainly driven by politicians and health administrators but should also be adopted by health professionals and the society as well (Drummond, Sculper, Torrance, O'Brien, & Stoddart, 2005). As so, any economic study in the field of medicine is “the comparative analysis of alternative courses of action in terms of both their costs and consequences” and wants to answer simultaneously two questions: which is the best medical option for a specific clinical outcome and what cost is the society willing to pay to obtain that outcome (Drummond, Sculper, Torrance, O'Brien, & Stoddart, 2005). They can consider the comparators similar in terms of clinical benefits and just address the cost problem (cost-analysis or cost-minimization analysis) or consider the 51 comparators with different clinical benefits and then compare the different costs involved (cost-effective analysis). When the clinical benefits, for instance life-years saved (LYS), are adjusted to utilities in terms of quality of life (QALY-quality adjusted life years) the studies are called cost-utility analyses and, when the clinical benefit is transformed into a monetary value the studies are called cost-benefit analyses. The representation of the simultaneous costs and effects of a clinical option can be plotted in a cost-effectiveness plane that will have four quadrants and the willingness to pay frontier defined by the society is represented by a line that crosses the centre of the axis and runs through quadrants I and III as shown in the figure 3: Figure 3 Cost-effectiveness plane 52 Legend: The cost-effectiveness plane is a simultaneous representation of the difference between two strategies in terms of both costs and effects (Drummond, Sculper, Torrance, O'Brien, & Stoddart, 2005). Costs and effects cross in zero and while the difference in costs changes from negative to positive from bottom to top, the difference in effects changes from negative to positive from left to right, providing four quadrants named by the corresponding Roman numbers I to IV. The willingness to pay frontier defined by the society is represented by a line that crosses the centre of the axis and runs through quadrants I and III. Cost-effective strategies are placed in quadrants I and III below the willingness to pay frontier or in quadrant II. ΔDifference between strategies in terms of costs or effects. √- Cost-effective strategies. In general, the alternative or option in study when compared with the present clinical practice will imply a higher cost and the difference has to be quantified and compared to the threshold defined by the society. When plotting effects and cost in the costeffectiveness plane the option in study usually comes on the right upper side (quadrant I) by providing more effects but at a higher expense (Black, 1990). Cost-effective strategies should be below the willingness to pay threshold, usually in the quadrant I as already mentioned. In very rare situations the new option will provide more effects with a lower cost (quadrant II). Other rare situations might be below the willingness to pay frontier in quadrant III but are discussible to the society as they will provide less effects although at a lower cost than the actual practice. The primary outcome measure for any economic evaluation is obtained by calculating the ratio between the difference in costs among strategies that are placed in the numerator and the respective difference in effectiveness that will be placed in the 53 denominator and this is called the incremental cost effectiveness ratio (ICER), as shown in figure 4 (Drummond, Sculper, Torrance, O'Brien, & Stoddart, 2005): Figure 4 Incremental Cost Effectiveness Ratio (ICER) Legend: The Incremental Cost Effectiveness Ratio is the calculation of a ratio among differences in costs between two strategies versus the difference in effects of the same strategies (Drummond, Sculper, Torrance, O'Brien, & Stoddart, 2005). Costs need to be in the same monetary currency and effects in the same clinical unit, like Life Years Saved (LYS) or Quality Adjusted Life Years (QALY). The first option in each part of the equation is the strategy in study (the study hypothesis) and the second option is the usual strategy in clinical practice (the current standard). ICERIncremental Cost Effectiveness Ratio. For diseases that develop for several years and that might impair not only survival but also the quality of that life as well, the standard effectiveness used are the years of life adjusted for they quality, namely the quality adjusted life years (QALY) (Weinstein & Stason, 1977), providing a cost-utility study instead of only a cost-effectiveness analysis. In this form of economic evaluation, the focus is on the quality of the health outcome produced and by incorporating the notion of value have a more broad applicability and are more useful to decision makers (Drummond, Sculper, Torrance, O'Brien, & Stoddart, 2005). Guidelines for the reporting of these studies have for many years recommended conducting cost-utility analysis, where the use of clinical benefits ICER = Cost new alternative – Cost current practice alternative Effect new alternative – Effect current practice alternative 54 should be adjusted to patient preferences (Russell, Gold, Siegel, Daniels, & Weinstein, 1996; Siegel, Weinstein, Russell, & Gold, 1996; Weinstein, Siegel, Gold, Kamlet, & Russell, 1996). Thus, utilities in terms of QALYs means that 1 year of live is multiplied by a utility factor between 1 and 0, providing different values for each single year of life depending on the quality of life of that year, resulting in a value that will vary between 1 QALY (one year with perfect quality of life) and 0 (death, by definition). These guidelines also suggest that preferences should be used by adopting population preferences instead of patient-reported values. Cost-utility analysis has the advantages in relation to cost-effectiveness analysis of using generic measures like QALY that potentiate comparability among studies, is a method to which various disparate outcomes can be combined into a single measure (QALY) and highlights consumer preferences or utilities, being the preferred method according to guidelines (Drummond & Jefferson, 1996). It is also the ideal method to compare cancer treatments as many of these treatments improve longevity and longterm quality of life but decrease quality of life during the treatment process itself (Drummond, Sculper, Torrance, O'Brien, & Stoddart, 2005). Preferences are an umbrella that describes the overall concept while utilities and values are different types of preferences, according to the following table as described by Drummond and colleagues (Drummond, Sculper, Torrance, O'Brien, & Stoddart, 2005) (Table 3): 55 Table 3 Preferences definitions and type of methods Question framing Response method Certainty (values) Uncertainty (utilities) Scaling 1 2 Rating scale Category scaling Visual analogue scale Ratio scale Choice 3 4 Time trade-off Standard gamble Paired comparison Equivalence Person trade-off Legend: Definition of type of preferences according to the options for responding (choice or scaling) and the framing of questions (with certainty or uncertainty), as suggested by Drummond and colleagues (Drummond, Sculper, Torrance, O'Brien, & Stoddart, 2005). In summary, options in sectors 1 and 3 measure values will those in sector 4 measure utilities. For the calculation of preferences, when using a cardinal measurement, the number must be attached to an outcome that in some sense represents the preference for that outcome in comparison to other and in theory in should be in an interval scale and equal intervals should mean the same difference along the scale (Drummond, Sculper, Torrance, O'Brien, & Stoddart, 2005). There is no best method and measuring preferences with most of these options are time consuming and complex. As so, in many cases, authors use pre-scored multiattribute health status classification systems such as the EQ-5D-5L from the EuroQol group as we did for the present thesis (EuroQol, 1990). The preferences scoring for this specific system was done using the time trade-off technique in around 3,000 United 56 Kingdom adults with the final scores falling in between 0 (death) and 1 (perfect health) (Dolan & Gudex, 1995; EuroQol, 1990; Herdman, Gudex et al., 2011). Our option on this EQ-5D-5L health status classification system is the result of several advantages: the instrument is established for a long time, is feasible, reliable and valid according to hundreds of published studies, it accommodates attributes that might be relevant for patients with gastric cancer or submitted to treatments for it but at the same time is very general allowing for application and comparison to “healthy” subjects, it is the recommended instrument by the National Institute for Clinical Excellence, it is not very time consuming, is easily applicable even by less instructed patients and finally is available in Portuguese language and was already used in Portuguese gastric cancer patients in the past allowing for comparison of results (Drummond, Sculper, Torrance, O'Brien, & Stoddart, 2005; EuroQol, 1990; Ravasco, Monteiro-Grillo, & Camilo, 2003). Although the EQ-5D-5L questionnaire still has no population norms published for the Portuguese population, the older but similar EQ-5D-3L system had only recently set the preferences for the general population using the time trade-off technique and also developed population norms, unfortunately both only after the development of the present thesis and associated studies (Ferreira, Ferreira, Pereira, & Oppe, 2014, 2014). Because currently there is no validated method to transform utilities from the EQ-5D3L to the EQ-5D-5L systems, we used the Spanish EQ-5D-5L utilities. From the available options Spanish utilities are the most similar, providing a Pearson’s correlation coefficient of r=0.946 for both EQ-5D-3L population norms (Ferreira, Ferreira, Pereira, & Oppe, 2014). 57 Data from several authors suggests that independently of the pre-scored multiattribute health status classification system used, health preferences measurements does not vary significantly as a function of demographic variables, including race, income, gender, population or country (Kaplan, 1994; Kaplan & Anderson, 1988; Wang, Furlong, Feeny, Torrance, & Barr, 2002). Also, some authors argue that utility-weighted QALY is a good approximation of utilities in most situations and a good basic definition of what is trying to be achieve in health care (Culyer & Wagstaff, 1993; Garber & Phelps, 1997). The willingness to pay frontier most widely used is 50,000 United States Dollars (USD) per LYS or per QALY and this value will allow for comparability among results of different models (Shillcutt, Walker, Goodman, & Mills, 2009). Other possibility for this controversial point could be the adoption of a threshold of twice the gross national income per capita as suggested by some institutions (Garber & Phelps, 1997; Sachs, 2001; Tan-Torres Edejer, Baltussen et al., 2003). Independently of the choice, the willingness to pay threshold should be further converted to the local currency of the modelled population (WorldBank, 2013). For Portugal, the option for the USD 50,000 would be 36,575 Euros (€) at a 2013 exchange rate and the option based on the gross national income per capita would be quite similar, USD 41,240 or € 30,433 after conversion. Finally, the analysis of the model has to be contextualized in terms of the perspective used, meaning that the costs that are incorporated might vary depending on the viewpoint adopted. Guidelines propose that the society perspective is the best option as it will incorporate all available costs independently of the professional group 64 radiotherapy or just best supportive care. The pre-scored multi-attribute health status classification system used was the Portuguese version of the EQ-5D-5L questionnaire and the reference test for the diagnosis was the gastroenterologist diagnosis, including the histopathology result when applicable. Finally, the aim of our forth study and also the final aim of this thesis were to determine the cost-utility of performing a 3 yearly endoscopic surveillance of high risk patients with extensive premalignant conditions versus no surveillance for the Portuguese population, an intermediate incidence population for gastric cancer. 65 Chapter IV - Publications 66 67 Screening for Gastric Cancer and Surveillance of Premalignant Lesions: a Systematic Review of Cost-Effectiveness Studies Areia M, Carvalho R, Cadime AT, Rocha Gonçalves F, Dinis-Ribeiro M Helicobacter 2013; 18: 325-37 Impact Factor: 3.511 (2012 Journal Citation Reports®, Thomson Reuters) Rank 19 of 74 in Gastroenterology and Hepatology 68 69 70 71 72 73 80 81 82 83 Prevalence of Gastric Precancerous Conditions: A Systematic Review and Meta-analysis Marques-Silva L, Areia M, Elvas L, Dinis-Ribeiro M Eur J Gastroenterol Hepatol 2014; 26:378–87 Impact Factor: 1.915 (2012 Journal Citation Reports®, Thomson Reuters) Rank 48 of 74 in Gastroenterology and Hepatology 84 85 86 87 88 89 96 97 98 99 100 101 102 103 104 105 Cost-utility analysis of endoscopic surveillance of patients with gastric premalignant conditions Areia M, Dinis-Ribeiro M, Rocha Gonçalves F Helicobacter 2014; Aug 28. doi: 10.1111/hel.12150 Impact Factor: 3.511 (2012 Journal Citation Reports®, Thomson Reuters) Rank 19 of 74 in Gastroenterology and Hepatology 112 113 114 115 116 117 118 119 Chapter V - Discussion 120 121 Improving survival of gastric cancer patients might be achieved by increasing the detection of lesions in an earlier stage and that goal could be accomplished by the surveillance of high-risk patients with extensive premalignant conditions at risk of progressing to invasive cancer. An endoscopic surveillance schedule every 3-years was suggested in recent guidelines on a clinical basis but better studies are needed to show its cost-effectiveness (Areia, Carvalho, Cadime, Rocha Goncalves, & Dinis-Ribeiro, 2013; Dinis-Ribeiro, Areia et al., 2012; Dinis-Ribeiro, Areia et al., 2012). The objective of the present thesis was to determine the cost-utility of this endoscopic surveillance every 3 years in patients with extensive gastric premalignant conditions compared with no surveillance. To accomplish that goal we first performed two systematic reviews and one cross-sectional study in order to obtain the best available clinical data for the Portuguese population and then a cost-utility economic analysis on this hypothesis. Preliminary studies With our first systematic review of studies on the subject of cost-effectiveness of gastric cancer screening or surveillance, we think to have accomplished the objective of obtaining the best available evidence published so far (Areia, Carvalho, Cadime, Rocha Goncalves, & Dinis-Ribeiro, 2013). That information was very helpful for the conception of our model and also by providing both strengths and weaknesses of the available evidence. 128 We think that by doing so we have accomplished a level 1b economic study according to the Oxford Levels of Evidence, by providing an “analysis based on clinically sensible costs or alternatives with a systematic review of the evidence and including multi-way sensitivity analyses” (Oxford, 2011). A comparison of our model with the available literature, which comprises only 3 published models whose results disagree, is presented in Table 5 (Dinis-Ribeiro, da Costa-Pereira, Lopes, & Moreira-Dias, 2007; Hassan, Zullo et al., 2010; Yeh, Hur, Kuntz, Ezzati, & Goldie, 2010). 129 Table 5 Published models of economic studies on endoscopic surveillance of premalignant conditions versus no surveillance, compared to the present thesis Study (1st author, publication year) Present thesis 2013 Dinis-Ribeiro 2007 Hassan 2010 Yeh 2010 Type of study Cost-utility Cost-utility Cost-effectiveness Cost-utility Country / Model Year Portugal 2013 Portugal 2005 USA 2008 USA 2007 Intervention Endoscopy every 3 years Chromoendoscopy every 1 year Endoscopy every 1 year Endoscopy every 10 years Perspective Societal Health service Health service Societal Threshold / Unit USD 50,000 (36,575 Euros) Euros 50,000 USD 100,000 USD 50,000 Population Extensive premalignant lesions Extensive premalignant lesions Intestinal metaplasia Intestinal metaplasia Sensitivity analysis One-way, Probabilistic One-way Two-way, Probabilistic One-way, Probabilistic Source of Effectiveness LYS, QALY QALY LYS QALY Result (ICER) € 18,336 per QALY € 1,868 per QALY USD 72,519 per LYS USD 544,500 per QALY Relevant variables in sensitivity analysis Utilities Quality of life after surgery Cancer incidence reduction Surgical risks Gender proportion Chemotherapy cost Cancer downstaging Success of endoscopic resection Strengths / Limitations Strengths: Limitations: Limitations: Limitations: Type of review Systematic review Not detailed Not detailed Not detailed Model complexity Extensive Simple Simple Simple Number of variables (clinical, cost) Extensive (58, 38) Small (3, 6) Small (7, 2) Moderate (25, 10) Utilities Utilities from specific population Utilities from other populations No utilities Utility valuations not detailed Threshold According to guidelines Not detailed Not conventional Conclusion Endoscopy every 3 years is costeffective for extensive atrophy or intestinal metaplasia Chromoendoscopy every 1 year is cost-effective for extensive atrophy or intestinal metaplasia Endoscopy every 1 year is costeffective for intestinal metaplasia Endoscopy every 10 years is not cost-effective for intestinal metaplasia Legend: USAUnited States of America, LYSLife years saved, QALYQuality adjusted Life years, ICERIncremental cost effectiveness ratio, USDUnited States Dollars. 130 There are, however, some limitations in the present model. Although it is extensive, it does not definitely accommodate all possible clinical real life options, and it is impossible to know how far or how near we are to the perfect model. Also, even by using utilities obtained directly from the studied population, the utilities valuation is also open to bias and the confidence intervals available are wide (Areia, Alves et al., 2014). Finally, the ranges and distributions for sensitivity analysis are not generally available in the literature and the approximations we used might not be the best option. 131 Chapter VI - Conclusion and Further Research 132 133 Conclusion In conclusion, according to our model, endoscopic surveillance of patients with premalignant conditions such as extensive atrophy or intestinal metaplasia, every 3 years, in an intermediate-risk country such as Portugal, is cost-effective. This strategy provides an incremental cost effectiveness ratio of only € 18,336 per QALY, well below the usual willingness-to-pay threshold of € 36,575 (equivalent to USD 50,000) and also below € 30,433 if the threshold was based on the gross national income, while surveillance at every 5 or 10 years were dominated as they provided less effectiveness at similar costs. Our model fitted real life estimates and proved to be robust in sensitivity analyses, as almost all variables evaluated in one-way deterministic analysis remained below the threshold and in Monte Carlo probabilistic analysis 78% of simulations would fall below the willingness-to-pay threshold. 134 Further research Further studies evaluating the impact of this option are needed to confirm if surveillance will in fact improve early gastric cancer detection rates, increase curative endoscopic resections, detect invasive cancer in earlier stages and ultimately lead to better overall survival for gastric cancer patients. The hypothesis modelled in the present thesis should be replicated in other countries in order to confirm or not its cost-utility in other populations, particularly in countries with only low to moderate gastric cancer incidence. Moreover, effectiveness could be further studied in terms of monetary benefits by transforming health benefits in monetary units and performing a cost-benefit analysis, along the cost-utility analysis. Finally, screening instead of surveillance could be modelled in a similar manner, to decide if in a specific population it would be cost-effective to screen instead of surveillance and in which specific high-risk population could this strategy be offered. 135 References 136 137 Abrams, J. A., & Wang, T. C. (2010). Adenocarcinoma and Other Tumors of the Stomach. In M. Feldman, L. S. Friedman & L. J. Brandt (Eds.), Sleisenger and Fordtran's Gastrointestinal and Liver Disease: Pathophysiology, Diagnosis, Management (9th ed., Vol. 1, pp. 887908). Philadelphia, USA: Saunders Elsevier. Areia, M., Alves, S., Brito, D., Cadime, A. T., Carvalho, R., Saraiva, S., . . . Dinis-Ribeiro, M. (2014). Health-related Quality of Life and Utilities in Gastric Premalignant Conditions and Malignant Lesions: a Multicentre Study in a High Prevalence Country. J Gastrointestin Liver Dis, 23(4), In press. Areia, M., Amaro, P., Dinis-Ribeiro, M., Cipriano, M. A., Marinho, C., Costa-Pereira, A., . . . Leitao, M. C. (2008). External validation of a classification for methylene blue magnification chromoendoscopy in premalignant gastric lesions. Gastrointest Endosc, 67(7), 1011-1018. doi: 10.1016/j.gie.2007.08.044 Areia, M., Amaro, P., Dinis-Ribeiro, M., Moreira-Dias, L., Romaozinho, J. M., Gouveia, H., & Leitao, M. C. (2008). Estimation of the extent of gastric intestinal metaplasia by methylene blue chromoendoscopy. Eur J Gastroenterol Hepatol, 20(9), 939-940. doi: 10.1097/MEG.0b013e3282f376ac Areia, M., Carvalho, R., Cadime, A. T., Rocha Goncalves, F., & Dinis-Ribeiro, M. (2013). Screening for gastric cancer and surveillance of premalignant lesions: a systematic review of cost-effectiveness studies. Helicobacter, 18(5), 325-337. doi: 10.1111/hel.12050 Areia, M., Dinis-Ribeiro, M., & Portuguese Society of Digestive Endoscopy, S. (2014). One day of upper gastrointestinal endoscopy in a southern European country. GE J Port Gastrenterol, 21(3), 97-101. Areia, M., Dinis-Ribeiro, M., & Rocha Goncalves, F. (2014). Cost-utility analysis of endoscopic surveillance of patients with gastric premalignant conditions. Helicobacter, In press. doi: 10.1111/hel.12150 Areia, M., Pimentel-Nunes, P., Marcos-Pinto, R., & Dinis-Ribeiro, M. (2013). Gastric cancer: an opportunity for prevention. Acta Med Port, 26(6), 627-629. Areia, M., Soares, M., & Dinis-Ribeiro, M. (2010). Quality reporting of endoscopic diagnostic studies in gastrointestinal journals: where do we stand on the use of the STARD and CONSORT statements? Endoscopy, 42(2), 138-147. doi: 10.1055/s-0029-1243846 Asaka, M., Sugiyama, T., Nobuta, A., Kato, M., Takeda, H., & Graham, D. Y. (2001). Atrophic gastritis and intestinal metaplasia in Japan: results of a large multicenter study. Helicobacter, 6(4), 294-299. Banatvala, N., Mayo, K., Megraud, F., Jennings, R., Deeks, J. J., & Feldman, R. A. (1993). The cohort effect and Helicobacter pylori. J Infect Dis, 168(1), 219-221. Black, W. C. (1990). The CE plane: a graphic representation of cost-effectiveness. Med Decis Making, 10(3), 212-214. Capelle, L. G., de Vries, A. C., Haringsma, J., Ter Borg, F., de Vries, R. A., Bruno, M. J., . . . Kuipers, E. J. (2010). The staging of gastritis with the OLGA system by using intestinal metaplasia as an accurate alternative for atrophic gastritis. Gastrointest Endosc, 71(7), 1150-1158. doi: 10.1016/j.gie.2009.12.029 Carneiro, F., Machado, J. C., David, L., Reis, C., Nogueira, A. M., & Sobrinho-Simoes, M. (2001). Current thoughts on the histopathogenesis of gastric cancer. Eur J Cancer Prev, 10(1), 101-102. Carneiro, F., Machado, J. C., Seruca, R., & Sobrinho-Simoes, M. (1999). E-cadherin changes in gastric carcimona. Histopathology, 35(5), 477-478. Carneiro, F., Oliveira, C., Suriano, G., & Seruca, R. (2008). Molecular pathology of familial gastric cancer, with an emphasis on hereditary diffuse gastric cancer. J Clin Pathol, 61(1), 25-30. doi: 10.1136/jcp.2006.043679 Caro, J. J., Briggs, A. H., Siebert, U., Kuntz, K. M., & Force, I.-S. M. G. R. P. T. (2012). Modeling good research practices--overview: a report of the ISPOR-SMDM Modeling Good