Economic evaluation in health research: cohort simultation and applications

ADVERTIMENT . La consulta d’aquesta tesi queda condicionada a l’acceptació de les següents condicions d'ús: La difusió d’aquesta tesi per mitjà del servei TDX (www.tesisenxarxa.net) ha estat autoritzada pels titulars dels drets de propietat intel·lectual únicament per a usos privats emmarcats en activitats d’investigació i docència. No s’autoritza la seva reproducció amb finalitats de lucre ni la seva difusió i posada a disposició des d’un lloc aliè al servei TDX. No s’autoritza la presentació del seu contingut en una finestra o marc aliè a TDX (framing). Aquesta reserva de drets afecta tant al resum de presentació de la tesi com als seus continguts. En la utilització o cita de parts de la tesi és obligat indicar el nom de la persona autora. ADVERTENCIA. La consulta de esta tesis queda condicionada a la aceptación de las siguientes condiciones de uso: La difusión de esta tesis por medio del servicio TDR (www.tesisenred.net) ha sido autorizada por los titulares de los derechos de propiedad intelectual únicamente para usos privados enmarcados en actividades de investigación y docencia. No se autoriza su reproducción con finalidades de lucro ni su difusión y puesta a disposición desde un sitio ajeno al servicio TDR. No se autoriza la presentación de su contenido en una ventana o marco ajeno a TDR (framing). Esta reserva de derechos afecta tanto al resumen de presentación de la tesis como a sus contenidos. En la utilización o cita de partes de la tesis es obligado indicar el nombre de la persona autora. WARNING. On having consulted this thesis you’re accepting the following use conditions: Spreading this thesis by the TDX (www.tesisenxarxa.net) service has been authorized by the titular of the intellectual property rights only for private uses placed in investigation and teaching activities. Reproduction with lucrative aims is not authorized neither its spreading and availability from a site foreign to the TDX service. Introducing its content in a window or frame foreign to the TDX service is not authorized (framing). This rights affect to the presentation summary of the thesis as well as to its contents. In the using or citation of parts of the thesis it’s obliged to indicate the name of the author 1 Statistics and Operations Research Doctoral Thesis, 2014 ECONOMIC EVALUATION IN HEALTH RESEARCH: COHORT SIMULATION AND APPLICATIONS Nuria Pérez Álvarez Thesis Directors: Guadalupe Gómez Melis, PhD Departament d’Estadística i Investigació Operativa Universitat Politècnica de Catalunya, Barcelona, Spain Roger Paredes i Deirós, MD, PhD Fundació de Lluita contra la Sida, Badalona, Spain Institut de Recerca de la Sida IrsiCaixa, Badalona, Spain Universitat Autònoma de Barcelona, Barcelona, Spain 2 3 This thesis was developed under the auspices of the Beca FIE, funded by the Generalitat de Catalunya. This program funds the companies that offer a project of research to be done by a PhD student in the frame of a PhD thesis development and the collaboration between the company and the university. Period: 03-2008 to 03-2011 Site of implementation: Fundació de Lluita Contra la SIDA and Universitat Politècnica de Catalunya. And partially supported by grants MTM2008-06747-C02-00 from the Ministerio de Ciencia e Innovación (period: 2009-2011) and MTM201238067-C02-01 from the Ministerio de Economía y Competitividad (period: 2013-2015). 4 5 6 7 ACKNOWLEDGEMENTS Para un investigador finalizar la tesis doctoral es comparable a recibir un óscar, es en este momento en el que se nos permite expresar públicamente una serie de agradecimientos que en otra situación podría considerarse cursi, e incluso fuera de lugar. Aprovechando la ocasión, debo y quiero agradecer a muchos/as que me hayáis acompañado en el proceso de “doctoración”. Además de conocimientos técnicos he madurado como persona, lo cual implica nuevas y mejoradas facetas y también malos hábitos y manías adquiridas, la mayoría de las veces va todo en el mismo pack e incluso se hace difícil discernir que es virtud y que es defecto. Tenemos la capacidad de ver, pero hay personas que nos enseñan a mirar, los que primero se hicieron cargo de esta labor fueron mis padres y hermana, a quienes debo mucho de lo que soy y de cómo soy. Hace 14 años que mi camino se cruzó con el de Lupe, desde entonces ha sido mi mentora y madrina en múltiples aspectos. Recuerdo mi primer congreso de estadística en Lleida y mi primera ponencia en Oviedo, reuniones y debates sobre metodología estadística y libros prestados, charlas animadas, baile, risas… Siempre presentes la generosidad y una gran calidad profesional y humana. Gracias por ayudarme a crecer, a veces dándome el empujón y otras veces dándome el espacio. A Ventura, que ya hace mucho que nos conocemos y, aunque él no lo diga, sé que le gusta la estadística. Participar en sus brainstormings es un random walk por la historia de la lucha contra el VIH/SIDA y por aquello que cambiará la historia de esta enfermedad. El inicio de esta tesis fue una pregunta clínica de Ventura y se ajustó siguiendo las directrices de Roger. Ambos tienen en común la capacidad de ver más allá identificando los puntos de “hot-research” y, 8 generalmente, todos a la vez. Roger, gracias por tu apoyo y por tener siempre la puerta abierta para mí. A Tom por mejorar el inglés de los apartados 1.1 y 1.2, él dice que la comunicación funcionaba y que sólo la mejoró; he intentado aprender el proceso de mejora para aplicarlo al resto de la memoria… confío en que no lo haya hecho mal del todo. A Analía, que me ha prestado su ayuda para resolver algunas dudas de gramática y semántica. También a Lupe y Roger que han hecho mucha parte de esta tarea. No quiero olvidarme de dar las gracias a los referees externos, cuyos comentarios han servido para mejorar este trabajo. Carme Macias, gracias por tu tiempo y tus ánimos. Haces la burocracia asociada a seguir un programa de doctorado muy llevadera. Lupe y Roger me regalaron una estancia en Harvard y la colaboración con Kenneth Freedberg y su equipo. Una experiencia increíble. Repetiría a pesar de haberla vivido con “frío peligroso” y en medio de la “explosión hibernal”. Esta vivencia me mostró el valor del tiempo: las reuniones de 15 minutos existen y son productivas. Además tuve la oportunidad de estar cerca de gente fantástica: Steve, Marvin, Alane, Binta y Ann. Con Ann he compartido charlas sobre docencia, doctorado, comunicación entre células, shopping en USA y gastronomía; los 2 últimos también los pusimos en práctica. He pertenecido al departamento de estadística e investigación operativa de la UPC como docente y como estudiante. Mis compañeros de docencia son un ejemplo de dedicación y generosidad, intento aprender de vosotros. Todos han tenido palabras amables e invitaciones a eventos sociales, a pesar de mi limitada convivencia con ellos. Entre los eventos sociales destaco las tardes boleras, derribando pinos; gracias por la diversión y los consejos técnicos compartidos: Toni, Hanna, Grevy y Jeaneth sois estupendos. 15 Unesco codes that describe the work done in the thesis: 120806 Markov Processes 120900 Statistics 120903 Data analysis 120904 Decision making procress (see 1207.06) 120912 Statistical association methods 120914 Statistical prediction methods 16 17 CONTENTS Page 1. Introduction 1 1.1. Background and motivation 3 1.2. Data sources 10 1.2.1. Data requirements 1.2.2. Data collection 1.2.3. Data generation 1.2.4. Some remarks on data sources 1.3. Techniques for economic evaluation 20 1.3.1. Definitions 1.3.2. The ICER, the cost-effectiveness p lane and the INB 1.3.3. Outline of the approaches used in health research 1.4. Goals and thesis structure 35 2. Treatment adherence p romotion strate g y in HIV infected patients. Decision trees 37 2.1. Decision trees 39 2.2. HIV infection and a promoting adherence program 41 2.2.1. Clinical background 2.2.2. Study characteristics 2.3. Results 44 2.4. Discussion and conclusion 46 3. Cost-effectiveness study for com p arison of bazedoxifene with raloxifene in osteoporosis prevention. Markov model 49 3.1. Markov model 51 3.2. Introduction to osteoporosis disease 61 3.3. Model to estimate the cost-effectiveness of bazedoxifene versus raloxifene 63 3.4. Results 65 3.5. Discussion and conclusion 69 4. HIV tro p ism testin g for maraviroc allocation. Markov model in n-stages 73 4.1. HIV antiretroviral treatment and HIV co-rece p tor usa g e test 75 4.2. Markov models in n-stages 78 4.3. Model definition for the HIV co-receptors testing 82 4.4. Results 87 4.5. Discussion and conclusion 94 18 5. Cost of a new treatment for the p aroxysmal nocturnal haemoglobinuria (PNH). Budget impact analysis 99 5.1. Budget impatct analysis (BIA) 102 5.2. The paroxysmal nocturnal hemoglobinuria 108 5.3. Model for assessing the PHN treatment 110 5.4. Results 113 5.5. Discussion and conclusion 120 6. DALYs in p ostmeno p ausal women with osteo p orosis. Health benefits quantification 123 6.1. DALYs calculation 126 6.2. DALYs calculation for p ostmeno p ausal women with osteoporosis using individual information 128 6.2.1. Data 6.2.2. Missing data study 6.2.3. Utility and disutility weight 6.2.4. YLLs and YLDs 6.2.5. Data analysis 6.3. Results 135 6.4. Discussion and conclusion 139 7. Conclusions 141 8. Future research questions 143 Annexes Annex I: Discrete-event simulation. Another techni q ue for model building 147 Annex II: Health outcome measures 149 Annex III: Ethical considerations in the use of the pharmacoeconomic studies 153 Annex IV: In p ut p arameters for ProAdh study in HIV infected patients 155 Annex V: Data analysis code for ProAdh study in HIV infected patients 159 Annex VI: Numerical exam p le for a cohort simulation usin g a Markov model 165 Annex VII: In p ut p arameters for bazedoxifene versus raloxifene model 167 Annex VIII: Analysis worksheets for bazedoxifene versus raloxifene model 177 Annex IX: Cohort simulation and cost-effectiveness analysis code with R 189 Annex X: Input parameters for the model of HIV tropism testing 209 Annex XI: Costs: Definitions and concepts 211 19 Annex XII: Input parameters for the PNH model 213 Annex XIII: Analysis worksheets for the PNH model 215 Annex XIV: Medical outcomes study (MOS) 221 Annex XV: Data analysis and DALYs calculation code with SPSS 223 Bibliography and References 233 Addenda Addendum I: Publications related with this thesis as of June 2014 253 Addendum II: Conference contributions related with this thesis as of June 2014 255 Addendum III: Other publications from September 2007 as of June 2014 257 20 21 LISTING OF FIGURES Figure 1.1. The components of an analytical model and the steps required to represent and assess a real system by means of modelling are displayed. Figure 1.2. The components of a simulation model and the steps required to represent and assess a real system by means of simulation are displayed. Figure 1.3. The cost-effectiveness plane. In the figure, the label and the decision about the compared treatments corresponding to each quadrant are indicated. Figure 2.1. Decision tree structure. This is a graphic representation of the context of a decision and its impact on health results. Figure 2.2. Chronogram of the study procedures by branch of health care intervention. Figure 2.3. Cost-effectiveness decision trees considering clinical variables in experimental and control groups. Figure 2.4. Cost-effectiveness decision trees considering health related quality of life variables in experimental and control groups. Figure 3.1. Representation of a simple Markov model. Figure 3.2. Decision tree structure for the Markov Model in Figure 3.1. Figure 3.3. Structure of a transition probabilities matrix for a three-state Markov model. Figure 3.4. Graphic representation of the simulation model. Figure 3.5. Cost efficacy plot for the two evaluated treatments. Figure 3.6. Cost-effectiveness acceptability curves: bazedoxifene versus raloxifene. Figure 3.7. Cost-effectiveness of bazedoxifene versus raloxifene in postmenopausal osteoporotic women. 22 Figure 4.1. Description of the available tropism tests grouped by phenotypic and genotypic procedure. Their characteristics and their accuracy in virus detection are reported. Figure 4.2. Representation of a 2-stage Markov model. Figure 4.3. Flow chart for the tropism test result and treatment allocation. Figure 4.4. Graphic representation of the different health states included in the Markov model. Figure 4.5. Graphic representation of the percentage of individuals achieving indetectability over time, results from the MERIT-ES study. Two phases were distinguished: from week 0 to 24 and from 24 to 48 weeks-depicted using the dashed line. Figure 4.6. Cost efficacy plot for the three therapeutic strategies evaluated. The values for cost and utility for a year of simulation are displayed. Figure 4.7. Cost-effectiveness plane for the comparison between therapeutic strategies. Figure 4.8. Cost-effectiveness acceptability curve for the three therapeutic strategies by different values of the ceiling ratio. Figure 5.1. Schematic representations of the budget impact and ICER computation. Both approaches compare the current environment with a new one –which can include a new therapeutic strategy-. BIA reflects resources use and the ICER considers both costs and health outcome. Figure 5.2. Direct and indirect total costs for the eculizumab treatment and the standard of care. Figure 5.3. Diagram indicating the direct and indirect costs for the different scenarios of the sensitivity analysis. 23 Figure 5.4. Budget impact of the use of eculizumab versus the standard of care for 5 years while the percentage of patients treated with eculizumab increases over time. Figure 6.1. Mean (95% CI) DALYs loss, undiscounted and discounted per group and for the total population. Figure 6.2. Mean (95% CI) DALYs loss, undiscounted and discounted per group and for the total population < 65 years of age and ≥ 65 years. 24 LISTING OF TABLES Table 1.1. Summary of the techniques of analysis to jointly evaluate health outcome and economical cost. The health outcome units and the calculation required to compare between therapies are characterized. Table 1.2. In the comparison of two treatments, 4 situations are possible according to cost (rows) and effect (columns). Table 3.1. Two-cycle Markov trace for a 3-state Markov model with health states: Well, Sick and Dead. Table 3.2. Total Cost, Incremental Costs, QALY, QALYs Gained, and ICER Table 4.1. Sensitivity, specificity of the assessed co-receptor tests are displayed. Table 4.2. Fifteen scenarios were created changing the input parameters to perform a sensitivity analysis. The Scenarios 1 to 8 are one-way analysis, and the following ones are two-way analysis. Table 4.3. Results of the cost-effectiveness study, in terms of costeffectiveness ratios (ICERs), according to the various possible scenarios. Table 5.1. Summary of direct, indirect and total costs per patient year by treatment group. Table 5.2. Direct and indirect costs are displayed by the different scenarios and by treatment group. Table 5.3. Budget impact analysis of the use of eculizumab versus the standard of care for 5 years. Table 6.1. Socio-demographics, clinical characteristics and participant background for all included women and by study group. Table 6.2. Mean (95% CI) DALYs loss, undiscounted and discounted by the presence of a risk factor. Table 6.3. Variables associated with DALY loss, undiscounted and discounted; adjusted by BD and previous osteoporosis BF. 7 care services are traditionally the primary source of social expenditure and have been at the core of many measures aimed at reducing costs and increasing efficiency. Some European countries such as Greece, Ireland, Italy, Portugal, and Spain, have reduced healthcare spending and introduced low ceilings on increases in the healthcare budget. Other countries have reduced the operational costs of health services and the prices paid to providers for goods, services, and tangible assets*. Some of these measures directly affect users in terms of payment for treatment, visits, hospitalization, and access to health technologies and drugs**, 8, 9. Consequently, rationalization of available resources and selection of the most beneficial and sustainable therapeutic strategies have become a priority10. It is necessary to evaluate the costs and benefits of available therapeutic strategies when attempting to make major improvements in health care. Indeed, decisions about public health and health care delivery increasingly rely on studies that assess the cost-effectiveness of medical services11. It is clear that a standardized set of methodological tools should be developed. The International Society for Pharmacoeconomics and Outcomes Research (ISPOR) provides a series of healthcare-specific economic concepts and facilitates a forum for discussion and guidelines for the development of research on healthcare costs and outcomes12. Pharmacoeconomics is the scientific discipline that evaluates the clinical, economic, and humanistic aspects of health care interventions in order to provide health care decision makers, providers, and patients with valuable information for allocating resources and obtaining optimal *Austria,Belgium,theCzechRepublic,Denmark,Estonia,Greece,Ireland,theNetherlands,Portugal, Spain,Slovenia,andtheUnitedKingdom.  **TheCzechRepublic,Denmark,Estonia,France,Greece,Ireland,Italy,theNetherlands,Portugal, Switzerland–andalsotheprivatehealthinsuranceintheUnitedStates‐raisedusercharges. 8 outcomes. The health care interventions include pharmaceutical products, diagnostic tools, services, programs, and activities to promote, generate, or re-establish health. Pharmacoeconomics incorporates and combines economics, clinical evaluations, risk analysis, health related quality of life, and epidemiology. It uses statistical and computer-based techniques to analyze drugs, medical devices, biotechnology, surgery, and disease-prevention services. In this context, the outcome and impact of different strategies can be examined by taking into account cost and health gain in order to address the following questions:  Which health care interventions should be included in the clinical care guidelines for a particular disease?  Which is the best health care intervention for a particular subset of patients?  Which is the cost per unit of outcome for a concrete health care intervention?  Will patient health related quality of life be improved by applying a particular health care intervention? The general aim of this thesis is to assess several methods to answer real clinical questions related to health resources rationalization. An overview of techniques and illustrations on the evaluation of cost and health outcomes to compare different strategies are presented and discussed. The Spanish health care system The data discussed here apply to the Spanish population, its health care system, and its expenditure on health care. In Spain, life expectancy at birth increased by more than two years from 1995 to 2005 and now stands at 80.23 years (76.96 for men and 83.48 for women)13. Application of various techniques to project life 9 expectancy in Spain for 2050 reveal values of 81 to 85.38 years for men and 87 to 91.97 years for women13-15. It is important to note that all the estimations in the studies cited indicated an increase in life expectancy. The increase in survival is linked to a larger number of citizens affected by a chronic disease. At least one over six Spanish adults (15 years old and older) suffers one of them. The lumbar pain (18.6%), arterial hypertension (18.5%), arthroses, arthritis and rheumatism (18.3%), high cholesterol (16.4%) and cervical pain (15.9%) are the most common16. A study published on 2002 reported that the Spanish population older than 65 years old suffers a mean of 1.8 chronic diseases (Standard Deviation=1.2, Minimum=0, Maximum=5). Being the hypertension the most prevalent (40.1%), followed by osteoarticular (24.0%) and cardiovascular diseases (18.4%) and sight impairment (16.6%)17. In order to gain a perspective of the impact of disease on the cost of the Spanish Health Care System, we analyzed the following 5 groups of diseases: HIV/AIDS, cancer, respiratory diseases, cardiovascular diseases, and neurological diseases. For each group, we used various sources to obtain information on the number of cases and the corresponding costs. The information provided below must be interpreted with caution, as the sources are not homogeneous. Neurological diseases generate the highest costs for the health care system. During 2004, between 6 and 7.5 million people had a neurological disease; the cost of treatment to the health system was over €10.8 billion*,18. This group was followed by cardiovascular diseases, which generated an expenditure of more than €9 billion per year19. Cardiovascular diseases have a high impact on mortality and cause 31.7% of deaths20. *Billionisequivalenttoaonethousandmillion,i.e.,109. 10 As for respiratory diseases, asthma affects 2.5 million people and costs €1.48 billion, and chronic obstructive pulmonary disease affects 4.8 million people and costs €3 billion21-23. Every year 162,000 of new cases of cancer—excluding skin cancer and melanoma—are diagnosed in Spain. In 2003, these generated costs of about €1.75 billion (colorectal, breast, prostate, and uterine cancer)24, 25. Treatment of HIV/AIDS affects fewer people and costs less than the other disease groups. In 2005, the Spanish health care system estimated the cost of antiretroviral drugs to be €0.423 billion (i.e., 423 million)26. 1.2. Data sources Several considerations should be taken into account in an economic evaluation. Data (information on cost and health gain per treatment strategy) are collected from real sources or generated through simulation. The type of the health-economic evaluation performed and the input data required depend on the definitions selected for cost and health outcome. The information used in the economic evaluation is detailed in subsection 1.2.1. A description of the available methods for obtaining data, namely, by real data collection or analytical models and simulation, is given in subsections 1.2.2 and 1.2.3, respectively. The Section ends with a discussion of the features of the various data sources. 1.2.1. Data requirements For a model to enable rationalization of resources and thus produce the largest possible gain in health per monetary unit, a series of points must be taken into account. These include the population of interest, the characteristics of the study cohort, the course of the disease under 11 study, available treatments or health care interventions, treatment efficacy*, adverse events, and cost. The disease of interest and the cohort characteristics depend on the target population, which can be a subset of the patients affected by the disease. It is necessary to gain knowledge of disease course, incidence, and guidelines for diagnosis and treatment. The intervention to treat the health problem studied can be a combination of tests, care services, and drugs, and it is of interest to analyze both their efficacy and their side effects. Efficacy should be measured objectively, i.e., it can be expressed in terms of enhanced health related quality of life and selfsufficiency, number of clinical events avoided, number of patients without treatment failure, or even health gain expressed as a monetary value. Apart from disease course and therapeutic options, the model should include the associated costs of each health care intervention, which are expressed in monetary terms. The items and services that are included in the cost calculation should be stated in the study plan (e.g., drugs, health care, and patient’s travelling expenses). The effectiveness (or efficacy) measure chosen and the list of items included in the costs of therapy define the type of the pharmacoeconomic study (described in subsection 1.3.1). Once the therapeutic strategies to be compared have been decided and the terms of the comparison are made, the data collection or generation process is designed. * ThetermsEfficacyandEffectivenessaregoingtobeusedwiththefollowingmeaning:Efficacyentails howadrugperformsinanidealorcontrolledcircumstance,asinthecontextofaclinicaltrial.However, effectivenessdescribesadrug'ssuccessinreal‐worldcircumstanceswherethepatientpopulationand othervariablescannotbecontrolled,i.e.underusualcircumstancesofhealthcarepractice. [REFERENCE:http://www.contextmattersinc.com/use‐of‐efficacy‐and‐effectiveness‐often‐misleading‐ and‐may‐skew‐reimbursement‐decisions‐presented‐at‐ispor‐europe‐2012/] 12 1.2.2. Data collection The data needed to build a pharmacoeconomic model can come from prospective randomized clinical trials (RCTs) and observational studies, retrospective databases or clinical files, expert panels (expert opinion), patient surveys, published literature, treatment guidelines, and research institute databases, such as those of the World Health Organization27, the Statistical Office of the European Communities15, the Instituto Nacional de Estadística28, or the Institut d’Estadística de Catalunya29. The availability, advantages, and disadvantages of the different types of databases are discussed below. 1.2.2.1. Prospective RCTs and observational studies In prospective RCTs, patients are randomly allocated to the intervention of interest and followed up for a defined period of time. The main goal is to evaluate and compare the health outcome, usually efficacy and/or safety, of the intervention. Within this framework, monetary cost can be easily registered to perform a pharmacoeconomic evaluation30, 31, but the utility of the data generated is limited owing to homogeneity in patient characteristics, fixed screening and follow-up schedules, and finalization of data collection when patients discontinue the study treatments. The design of RCTs makes it difficult to evaluate the therapeutic strategies for a large variety of patients, since it reduces the chance of unexpected outpatient visits and the need for symptom-driven diagnostic procedures and implies a lack of information when the patient discontinues the study. These restrictions prevent an extrapolation of health cost results for patients in daily clinical practice. This limitation is especially important in the field of prevention and chronic maintenance therapy. Observational studies assess patients with similar characteristics who differ with respect to the specific factors under study; the health 13 interventions are not controlled. During the follow-up, changes in the available therapeutic strategies and guidelines for treatment of a disease can vary and invalidate future conclusions, thus leaving them outdated. Observational studies have fewer limitations than the pharmacoeconomic studies associated with clinical trials; however, both are time-consuming and involve cost expenditure. Decision making in healthcare usually requires rapid access to information. The data from previous RCTs and observational studies can be used, but they have the same drawbacks as described above, except for the time and money consumed to obtain results. 1.2.2.2. Retrospective databases and clinical files Retrospective data analysis measures effectiveness and can provide “real-world” data. Cost-effectiveness analyses based on retrospective databases or clinical files can provide real-time, relevant, and comprehensive decision-making tools. Retrospective analyses are quick and relatively inexpensive to perform. They reflect specific populations that cannot be easily studied using RCTs. Retrospective databases tend to cover more realistic time frames, since they are not constrained by the limitations of a set trial period. Existing databases can provide a set of variables, and analyses of these data can reveal real-world prescribing patterns. The disadvantages of retrospective database analyses used for economic evaluations include the fact that some of the study variables are not directly precisely recorded32, which reduces the quality of the information. 1.2.2.3. Expert panels The increase in the number of pharmacoeconomic studies in the last decade has promoted the development of guidelines for the conduct of economic evaluations in many countries. It is noteworthy that different 14 study designs can impact results. The use of expert judgement in decision analytic modelling is one area where design issues may influence the findings of a study33. Several researchers have suggested that expert judgement can be used successfully in pharmacoeconomic studies. Most acknowledge that expert opinion should be used as a last resort in pharmacoeconomic studies. Barr and Schumacher support its use when ideal data are not available and when, together with information from meta-analyses and other trial data, expert opinion can serve as a reasonable approximation34. Nuijten et al. also acknowledge the weaknesses inherent in the use of expert opinion, although they report that its application is not forbidden in modelling studies35. Evans suggests that the use of expert opinion need not be avoided as long as potential weaknesses are addressed and the techniques are applied appropriately36. Similarly, Halpern et al. recognise that expert opinion plays an important role in modelling studies but that it is subject to many errors and biases37. Expert opinions can be obtained by means of Delphi panels, modified Delphi panels, and round tables. The Delphi technique is a well-known method for consensus building based on a series of questionnaires delivered using multiple iterations to collect data from a panel of experts38. The areas of concern to be considered when obtaining expert opinion include the provision of baseline information or seed algorithms to panellists, the high attrition rate of panels, the criteria for selecting experts, and the definition of consensus 36. Despite the difficulties and limitations involved in this method, information gathered through an expert panel can cover the lack of appropriate information necessary to perform pharmacoeconomic studies. 15 1.2.2.4. Patient surveys Surveys have been used to understand the value that patients place on health care interventions. Understanding patient preferences can help to improve adherence and better predict the corresponding health outcomes. It is widely accepted that adherence is maximised when a treatment or intervention matches the patient’s preferences (World Health Organization [WHO]39, National Institute for Health and Clinical Excellence [NICE]40). Patient surveys have been used to demonstrate patient tradeoffs between treatment features and outcomes, and to quantify patient values. The approaches used are self-reported adherence and measures such as willingness to pay or maximum acceptable risk. The term “willingness to pay” is the maximum amount a person would be willing to pay, sacrifice, or exchange for a good. The parameter “maximum acceptable risk” was proposed by Johnson41. The objective of this approach is to estimate the maximum risk patients are willing to accept in order to achieve the therapeutic benefits of drug therapy. This maximum acceptable risk can then be compared against the actual or expected risk associated with a treatment to determine whether a treatment is acceptable to patients. 1.2.2.5. Published literature Published literature, treatment guidelines, and research institute databases can provide insight into disease outcome, patient characteristics, recommended treatment, disease prevalence, and treatment efficacy or effectiveness. 1.2.3. Data generation Analytical models or procedures based on simulations provide data without conducting real experimentation using patients. These methods 16 do not imply costs or danger in their performance, and the results are obtained relatively quickly. 1.2.3.1. Analytical models Analytical models are symbolic and yield general solutions to a problem. The model is constructed following specific rules written in terms of mathematical expressions which reflect as closely as possible the realworld problem. A general solution is obtained, and specific cases might be assessed by forcing the variables included in the model to take different values. Both the complication of constructing an analytical model for dealing with a complex problem and the oversimplification of the problem are limitations of this approach. The steps required to define an analytical model departing from a real system and applying the information obtained to solve it are presented in Figure 1.1. 23 how valuations should be made, which valuations should be used, and how the valuations of different individuals should be combined. The cost-benefit analysis measures the impact of an intervention on monetary units. The costs and also the benefits are assessed in monetary terms, for this reason it is necessary to set money values on health outcomes. The therapies are compared using the ratio Cost/Benefit. The advantage of this method is its simplicity when comparing between treatments; however, the difficulty in setting money values to health outcomes and the ethical issues related to a subjective quantification entail a scarce use of this type of study on health area. The cost-benefit analysis is mainly, but not exclusively, used to assess the value for money of very large private and public sector projects. This is because such projects tend to include costs and benefits that are less agreeable to being expressed in financial or monetary terms (e.g. environmental damage), as well as those that can be expressed in monetary terms. A small discussion on ethical issues related to the resources allocation is presented in Annex III. Discounting Discounting is a procedure that can be applied to all the previous analyses. In fact, the results of the measures described above should be reported indicating if they are discounted or non-discounted. Costs and health outcomes should be discounted to present values when they occur in the future, to reflect society’s rate of time preference. Accordingly, any costs or outcomes occurring beyond one year should be depreciated using standard methods. A common discount rate should be used to ensure the comparability of results across evaluations. The standard rate for the Reference Case is set at 5% per year. A rate of 0% should be used to show the impact of discounting and a 3% discount rate 24 must be used in a sensitivity analysis* for a comparison with published evaluations in other jurisdictions. The discount rates are expressed in real (constant value) terms, which are consistent with valuing resources in real (i.e., constant, inflation-adjusted) monetary units (Euros, dollars, etc.)48. Some countries have developed their own guidelines to perform pharmacoeconomic studies. One of the most complete and frequently used as a good example are the NICE guidelines49, where the suggestions are to apply a 3.5% of annual discount rate and vary the rate between 0% and 6% for the sensitivity analysis if results are potentially sensitive to the discount rate. The discounting rate applied in the model should be clearly stated in the results document. In the assessment of the therapeutic strategies by means of cost and health outcome, both cost and health measurement should be defined and calculated. The following table summarizes the terms of cost and health outcome that can be used to compare a health-care intervention. * Asensitivityanalysisconsistsinexaminingthechangesinresultswhentheassumptionsinthemodel arevaried.Generallyaneconomicevaluationisbasedonanumberofdebatablehypotheses, introducinganelementofuncertainty.Sensitivityanalysissuggestsvarytheinputparametersin differentwaystocalculateandevaluatetherobustnessoftheresultsunderdifferentassumptions. 25 Table 1.1. Summary of the techniques of analysis to jointly evaluate health outcome and economical cost. The health outcome units and the calculation required to compare between therapies are characterized. Method of analysis Health outcome measurement Terms used to compare between therapies Assumptions and comments Costminimization Any unit Therapies price Therapies compared have the same efficacy and tolerability Costeffectiveness Natural health units Cost/Effectiveness ratio The health effect quantification should be suitably selected for the assessed therapy Cost-utility Utility score such quality-adjusted life years (QALYs) Cost/Utility ratio Special case of the Costeffectiveness analysis. The score per health state can be debatable Cost-benefit Monetary units Cost/Benefit Ratio The assignation of monetary units to health states can be debatable Discounting Should be applied to cost and health outcomes involded in the described methods of analyses when future values should be discounted to present values. The rate of discount should be set up depending on the study aim. 1.3.2. The ICER, the cost-effectiveness plane and the INB Well known and widely used calculations and plots to display the results for the cost-effectiveness and cost-benefit analysis are the incremental cost-effectiveness ratio (ICER), the cost-effectiveness plane plot and the incremental net benefit (INB), which facilitate comparing the costs and benefits of new and existing health care interventions. 26 Incremental Cost-Effectiveness Ratio In the evaluation of treatments, either by pairs or all of them versus an established therapeutic standard of care (SOC), four situations can be identified in terms of cost and health outcome:  Cost treatment A<Cost treatment B; Effect treatment A>Effect treatment; Accept the treatment A, as it is both cheaper and more effective than B. It is a situation of dominance.  Cost treatment A>Cost treatment B; Effect treatment A<Effect treatment B; Reject the treatment A, as it is both more expensive and less effective than B. It is a situation of dominance.  Cost treatment A>Cost treatment B; Effect treatment A>Effect treatment; the magnitude of the additional cost of treatment A relative to the additional effectiveness should be considered.  Cost treatment A<Cost treatment B; Effect treatment A<Effect treatment; the magnitude of the cost saving of therapy A relative to its reduced effectiveness should be considered (See also Table 1.2). Table 1.2. In the comparison of two treatments, 4 situations are possible according to cost (rows) and effect (columns). In two of the combinations, the treatment selection is clear, the unclear ones are marked with a question mark. Cost A<Cost B A selected ? Cost A>Cost B ? B selected Effect A>Effect B Effect A<Effect B As a summary of the previous situations, the cost-effectiveness can be expressed as an incremental cost-effectiveness ratio (ICER) defined as the ratio of change in costs to the change in effects. 27 BtreatmenttmeasuremenEffectAtreatmenttmeasuremenEffect BtreatmentCostAtreatmentCost ICER    The ICER value might be considered as the monetary cost of the additional outcome caused by switching from treatment B practice to the treatment A. Assuming that the new treatment is more effective and its price is low enough, the new strategy is considered "cost-effective” or dominant. The ICER value can be directly compared to a pre-specified amount of money which represents the maximum cost health payers would invest to achieve one clinical benefit unit, and this value is defined as the willingness to pay (or ceiling ratio, Rc) benchmark. The advantage of ICER is that different interventions are evaluated in the same units and decision, between interventions, can be based on the cost/unit of result. Its drawback is that the ICER interpretation varies in function on the result of the difference between the effects and between the costs of the compared treatments. Also, there is a limitation on the confidence interval calculations, especially when the the effect of both treatments is close to the same measured value50. Cost-effectiveness plane The incremental cost-effectiveness plane represents the incremental cost and the incremental effect from a treatment A versus a treatment B as coordinates in a plot51. The plane is divided into four quadrants: the horizontal axis divides the plane according to the incremental cost (positive above, negative below) and the vertical axis divides the plane according to the incremental effect (positive to the right, negative to the left). The cost-effectiveness plane is presented in Figure 1.352. 28 NW “Dominated” NE “MayormaynotbeCE” Existingtreatment dominates Newtreatmentmore effectivebutmore costly   Newtreatmentcostly butlesseffective Newtreatment dominates “MayormaynotbeCE”“Dominant” SW  SE Figure 1.3. The cost-effectiveness plane. In the figure, the label and the decision about the compared treatments corresponding to each quadrant are indicated. NE = northeast quadrant; NW = northwest quadrant; SE = southeast quadrant; SW = southwest quadrant. Each quadrant has a different implication for the decision: i) If the ICER is calculated for the new treatment compared to the SOC and it falls in the southeast quadrant, with negative costs and positive effects, the new treatment would be claimed more effective (larger health gain) and less costly than SOC; in this case it can be said that the new treatment 'dominates' the SOC. Interventions falling in this 4th quadrant are always considered cost-effective. ii) If the ICER is located in the northwest quadrant, with positive costs and negative effects, the new treatment would be more costly and less effective than SOC (i.e., new treatment is 'dominated' by SOC). Interventions falling in the 2nd quadrant are never considered costeffective. iii) If the ICER falls in the northeast (or 1st) quadrant, with positive costs and positive effects, or the southwest (also named 3rd) quadrant, with negative costs and negative effects, trade-offs between costs and effects would need to be considered. The 1st and 3rd quadrants represent the situation where the new treatment may be cost-effective compared to SOC, depending upon the value at which the ICER is considered good Incremental Health Effect Incremental Cost 29 value for money i.e.: compared to the maximum amount that the payer is willing to pay for health effects. For instance, in the UK, the National Institute of Health and Clinical Excellence (NICE) at 2012 uses a threshold between £20,000–30,000 (€24,557-36,835, using the exchange ratio of 1£ = €1.23) per QALY gained when deciding which interventions to approve (interventions costing less than £20,000 (€24,557) per QALY gained are more likely to be approved than interventions costing more than £30,000 (€36,835) per QALY gained)53. In Spain an intervention costing less than €30,000 per QALY gained is considered cost-effective54 and the interventions in the range of €30,000-€45,000 per QALY are also susceptible to be labelled as cost-effective.55 When a treatment is not dominant, deliberations about the collateral potential benefits and costs gained or lost, in the context of the most efficient use of resources, can help in the election. When the ICER shows that the new treatment is less costly and more effective than the SOC the concern is to quantify the variability or uncertainty of this result. The ICER is usually calculated from point estimates of costs and effects without taking into consideration their variability. To account for this variability, sensitivity analyses changing the input parameters and probabilistic techniques to generate a range of input parameters can be used to generate a set of possible results which can be taken as a quantification of the uncertainty surrounding the estimates of costs and effects. There remains considerable debate concerning the presentation of joint uncertainty for estimates of cost-effectiveness. The calculation of confidence intervals can be complex. The possibility that the numerator and/or denominator tend to 0 complicate calculations even more. As a result of these challenges, a number of alternative methods for calculating confidence intervals have been proposed. These methods 30 include the use of Fieller's theorem and non-parametric bootstrapping56, 57. Incremental Net Benefit The incremental net benefit (INB) can be defined in terms of health gain, known as incremental net health benefit (INHB), or in monetary quantification becoming the incremental net monetary benefit (INMB). INHBs estimate a treatment's net clinical benefit after accounting for its cost increase versus an established SOC. Lynd58 has proposed a framework for calculating the incremental net health benefits (INHB) of different pharmaceutical treatments. Both benefits and adverse events associated with a treatment are quantified using available clinical trial or surveillance published data. A score reflecting the utility is assigned to each outcome in order to express all benefits and all risks in a common scale. The difference between the sum of the weighted benefits and the sum of the weighted risks of a treatment represent the net health benefits of the treatment. INHB is calculated as the difference between the NHB of the treatment of interest minus the NHB of an alternative treatment or the standard of care. A positive INHB indicates that the net benefits of treatment are larger than its competitor59. INMB would be defined analogously. 1.3.3. Outline of the approaches used in health research Techniques used for pharmacoeconomic evaluations performance in the health area and the ones susceptible to be adapted for application to our real case studies are described. According to the data source, the models were classified in dynamic models, Markov models and models based on real. The main methodological differences among studies were allocated in the data generation/collection; the provenance of the data on costs and 31 health outcome can be real or simulated using different methods. The data used in the assessment can be collected from clinical records or prospectively in the framework of a clinical study. Another difference is the numerical summaries and graphical displays chosen to be reported, they are based on the research question that should be answered. Some examples of models applied to several health area problems are described in the following. Dynamic models These models are useful for studying the nature of epidemics or disease trends over time. They are typically deterministic and non linear over time; they track the changing population and individuals constantly enter the model as they are born and exit the model as they die. The probabilities of suffering health events change with the time. They are difficult to implement and few works used them. Edmunds et al.60 used a dynamic model for assessing the cost-effectiveness of vaccination programmes on human papillomavirus (HPV). Markov models Markov models, also called health-state transition models, are widely used for cohort simulation. In this approach, the transition probabilities between health states do not change with time. In cancer research, Markov models are often used to simulate the disease evolution. A brief summary of three representative published works is here stated: Van de Velde et al.61 implemented a state transitions model to assess the effectiveness of HPV vaccine; considering the natural history of infection and disease, the probability of a woman of being tested for HPV and the life-long natural immunity. The aim of this study was to predict the impact of HPV-6/11/16/18 32 vaccination on the girl’s life time risk of HPV infection. Yang et al.62 used a Markov model to evaluate the cost-effectiveness of two available gene expression profiling test to learn about the breast cancer recurrence and guide the treatment. Two representative examples of cost-effectiveness studies on coronary heart diseases63, 64 used Markov models to simulate the health-states previous death. The outcomes measured included costs, life expectancy in quality-adjusted life-years (QALYs), incremental cost-effectiveness ratios, and events prevented. In the HIV/AIDS cost-health outcome studies, we identified three differentiated cohort simulation models: the implemented by Freedberg et al.65-68, the developed by Sanders et al.69 and the developed by Sax et al.70. These models have in common a state-transition model framework, but they are based on different clinical assumptions. Freedberg et al. have developed a mathematical simulation model of HIV disease, using the CD4 cell count and HIV RNA level as predictors of the progression of the disease. The input information used for modelling the course of the disease were the monthly probabilities of clinical events: changes in CD4 cell count, changes in HIV RNA level, development of opportunistic infections, adverse reactions to medications and death. A state-transition model framework was employed; wherein disease progression in a patient was characterized as a sequence of monthly transitions from one health state to another. Outcome measures included life expectancy, life expectancy adjusted for the health related quality of life - scale from 0.0 (death) to 1.0 (perfect health), lifetime direct medical costs, and cost-effectiveness in dollars per quality-adjusted year of life gained. 39 2. TREATMENT ADHERENCE PROMOTION STRATEGY IN HIV INFECTED PATIENTS. DECISION TREES Decision trees are used to describe the possible choices and their consequences, in terms of the health outcome and resources expenditure. This method was used to assess an HIV antiretroviral treatment adherence program. The comparison between the intervention group and the standard of care is performed in terms of costeffectiveness and using real data collected in the framework of a clinical trial. 2.1. Decision trees A decision tree (or tree diagram) is a decision support tool that uses a graph of available options and their possible consequences, including chance event outcomes, resource costs, and utility. The branches off the initial decision node represent all the therapeutic strategies that are to be compared. A series of probability nodes of each strategy branch can be used to reflect uncertain events, usually within a relatively short time frame. The outcomes at the end of each pathway are values that reflect both the cost and the health effect associated with that pathway. Usually, the outcomes are grouped into health states which are characterized by a utility measure and a monetary measure of cost84. Example: The figure 2.1 is a graphic representation of the context of a decision and its impact on health results. In this case, the potential outcomes are Well, Sick and Dead, which should be defined in such a way that they are exhaustive, but 40 exclusive, i.e., they cover all possible outcomes, but a patient cannot be in more than one state at a time. The available therapeutic strategies to treat the health problem are A and B. The probabilities of achieving a health outcome for the studied treatments are known and displayed in the diagram (p and q). A logical constraint in the final nodes for the possible health outcomes is that the sum of the probabilities must be 1 . Figure 2.1. Decision tree structure. This is a graphic representation of the context of a decision and its impact on health results. A and B can be used as a treatment for the health problem assessed. The potential outcomes are Well, Sick and Dead. p and q are the probabilities of achieving the fist two health states, 1-p-q is the probability of Death. The square indicates a decision node, the circles represent the probability of the event, and the triangles indicate a final state. Health problem Treat. A Treat. B pA qA 1-pA-qA Well Sick Dead Well Sick pB qB 1-pB-qB Dead 41 Cost-effectiveness (CE) and incremental cost-effectiveness ratio (ICER) can be calculated to compare outcomes between groups. The CE and ICER are calculated at the end of the follow-up as described in subsection 1.3.2, CE is calculated as the cost divided by efficacy and ICER is the difference of cost between the treatment A and treatment B divided by their difference in effects. The decision tree offers a static portrait of a dynamic process. It is relatively easy to construct and use this approach, although it only works for micro-circumstances (i.e., well defined systems, described by few and well characterized features and usually in a bounded time), where the information does not come from different studies or populations, and it is not necessary to adjust for factors. Furthermore, duration of follow-up should be the same for all patients and branches. The difficulty to represent a disease characterized by the repetition of events in the time (such as chronic diseases: complications, recurrence and progression) and the impossibility to assign utility values to the health states and a discount rate to the costs are limitations of this approach. 2.2. HIV infection and a promoting adherence program 2.2.1. Clinical background HIV infection continues to be a major health epidemic problem. The World Health Organization (WHO) estimated that there were 34.0 million [31.4 million–35.9 million] people living with HIV worldwide at the end of 2011. In 2011, an estimated 2.5 million [2.2 million–2.8 million] new HIV infections occurred and 1.7 million [1.5 million–1.9 million] annual deaths were due to AIDS77. The WHO estimated the number of people living with HIV in Spain, among adults aged 15 years and older, to be 150,000 42 [130,000-160,000], and the prevalence in this setting was 0.4%[0.40.5]78. Significant advances in antiretroviral treatment have been made since the introduction of zidovudine (AZT) in 1987. With the advent of highly active antiretroviral therapy (HAART), HIV-1 infection is now manageable as a chronic disease in patients who have access to medication and who achieve durable virologic suppression79. Accessibility to antiretroviral therapies is general because the Spanish health care system provides universal health care free of charge for the patients. Nowadays concern is whether patients take the prescribed medication, as well as they follow the treatment dosage. Poor adherence to combined antiretroviral therapy (cART) has been shown to be an important determinant of virologic failure, emergence of drug resistant virus, disease progression, hospitalizations, mortality, and, consequently health care costs. The challenge is to achieve a high long term adherence and break the barriers to optimal adherence. The obstacles to overcome may be from individual (biological, socio-cultural, behavioural), pharmacological, and societal factors80. 2.2.2. Study characteristics A program to promote adherence in HIV naïve patients that start cART was established. The experimental group received the standard care of treatment and a psychoeducational adherence-based intervention consisting in 3 sessions of 1 hour of duration each. The visits were performed in the moment of cART starting, 2 weeks and 4 weeks later. During these sessions the beliefs of the patient about the HIV disease and his/her circumstances that prevent the patient to be adherent to the cART, including conceptual, behavioural and motivational areas, were discussed and the importance of the adherence was emphasized. The control group did not participate in the psychoeducational adherence- 43 based intervention program. See the chronogram of the study in figure 2.2. Experimental Group       0 2 4 12 24 36 48 PSABI PSABI PSABI BBT MA MA MA MA CV CV CV CV CV QoLA QoLA QoLA  Control Group       0 2 4 12 24 36 48 BBT MA MA MA MA CV CV CV CV CV QoLA QoLA QoLA Figure 2.2. Chronogram of the study procedures by branch of health care intervention. Time expressed in weeks (w) of follow-up (48w, considered equivalent to 1 year). PSABI is the psychoeducational adherence-based intervention. BBT is the Baseline blood test, CV is the Clinical visit, MA is the Monitoring analysis and QoLA is Health related quality of life questionnaire assessment. The performance of the program was evaluated in terms of costeffectiveness for different health outcomes. Data was collected through a prospective clinical trial designed to evaluate the health outcome in terms of HIV RNA viral load, CD4 cells count and health related quality of life variables at 1 year of follow-up. Forty treatment-naïve participants were randomized to the experimental and control groups. Clinical, economical and health related quality of life variables were assessed from the RCT data base and the direct cost of the hospital medical supplies. The numerical variables were expressed as mean (Standard deviation, SD) or as median and interquartile range (IQR) and compared using the t or Mann-Whitney test. For the categorical variables, percentages and/or number of patients were given and compared using the χ 2 or Fisher exact test (as appropriate). Further Time (weeks) Time (weeks) 44 detail on cost and effectiveness input data and calculations are provided in Annex IV and Annex V. 2.3. Results Participants were all men with a median (IQR) of 35 (30-45) years old, who were infected mainly through sex with other men (90%). The median number of cART changes during the study was 2, with a minimum of 0 and a maximum of 4 changes. Initially, 20 patients were allocated in each treatment group but 5 and 2 were lost to follow up in the control and experimental groups, respectively. To assess both cost and the clinical and health related quality of life outcomes of interest six decision tree models were built. The first two present the results for viremia control and the immune recovery, the next ones reflect the quality of live improvements (figure 2.3 and 2.4). These models compared the performance of the patients attending the adherence program with the individuals receiving the standard of care. The mean (SD) cost per patient month (PPM) was €1,252 (460) in the experimental group and €1,139 (275) in the control group. The percentage of patients that reached the end of the study with virological suppression was larger in the experimental group (94.4% versus 86.7%; not statistically different, p-value=0.579). The CE indicates that the cost per 1% more of patients with virological response is slightly larger in the experimental group (€1,326 PPM versus €1,314 PPM). The ICER indicates that, per 1% additional in viral suppression outcome, the incremental cost is €14.53 PPM. The percentage of individuals that show an improvement of 100 or more CD4 cells/mm3 was larger in the control group (80% versus 72.2%; not statistically different, p-value=0.699). 45 When this outcome is assessed, the adherence program is not costeffective (Figure 2.3). Figure 2.3. Cost-effectiveness decision trees considering clinical variables in experimental (EG) and control groups (CG). The ICER for the Undetectable Viral load marker is €14.53 and €-14.53 for the CD4 change as an incremental cost per 1% of increasement in the health outcome. Decision is represented by the square, the circle is a chance node, and triangle represents a final node. Mental and psychological, and global health scores were favourable to the experimental group in comparison with the control group; although the differences between groups were not statistically significant for any of the scores. Considering the cost added for the adherence promotion visits, the minimum cost therapeutic strategy should be chosen in this population. The percentage of every outcome and the CE ratios are displayed in Figure 2.4. 46 Figure 2.4. Cost-effectiveness decision trees considering health related quality of life variables in experimental (EG) and control groups (CG). The ICERs for the Physical health, Mental and psychological health, social relationships and global health are €- 11.30, €25.42, €-20.34 and €101.70 respectively. Decision is represented by the square, the circle is a chance node, and triangle represents a final node. 2.4. Discussion In our study there were no significant differences in the health outcomes between control and experimental programs. In terms of the trend found on the descriptive analysis it can be said that the patients in the psychoeducational adherence program had a scarce benefit in terms of achieving undetectable HIV viral load, compared with the patients in the control group. The HIV Unit where the trial was performed stresses the need of educating and making the patients aware of the treatment 47 adherence importance; the standard of care include interventions to help patients to understand the HIV infection, the drugs role and connect these terms with their routine and believings. This might be due to the fact that the adherence in our patients is greater than 80%. Then, it can be supposed that the standard of care in terms of adherence sensitivity is greater than in generalized clinical practice. If the program is implemented in units of care without specific interventions to help the patient to deal with the disease, a larger improvement in adherence, and consequently, in the health outcomes can be expected. Increasing the follow-up would be valuable to quantify the changes in immunological and health related quality of life scores and characterize the program effects in the long term. The health resources used were registered using the clinical files and some information on visits and prescribed drugs done out of the HIV unit could be ignored in our register. We assume that undereporting of resources used was balanced between both groups, and this did not significatively affect to our results. In spite of the study limitations and the lack of generalization of the results to the HIV infected patients visited in other clinical units, it was an asset to manage the intern available resources in the unit where the study was performed. The conclusions of this work are similar to what Goldie et al.65 reported, where they mention that in spite of improving the patients’ health related quality of life “ the cost of the programme represented a key variable” . The decision trees are very useful to describe situations where a simple choice and the set of possible outcomes are not very extense. The simplicity of this technique has the limitation of not reflecting the evolution of the health outcomes over time, in this method the value at the end of follow-up is used as an indicator of success or failure. 48 Conclusion The study performed can guide the selection of the therapeutic strategies applied to the clinical practice. The program to promote the HIV treatment adherence resulted in a few immunological or health related quality of life improvement, it seems cost-effective in terms of virological suppression if the decision-makers on health resources allocation consider worthy increasing the treatment cost. In our setting the increase estimated is of €14.53PPM to obtain a 1% of additional health outcome. 55 is the “residual probability” computed as the difference from 1 of p WS +p WD (see figure 3.3). This method allows obtaining information that is not usually reported in published health studies (research papers, epidemiological tables, etc.). The model is initially filled by distributing the simulated individuals across a number of starting health states according to parameters defining the probability of being in each of these states. These parameters can be extrapolated from the sickness prevalence in the population of interest. This is done by specifying the dimension of the set of states, which is 1×s, where s is the total number of health states in the Markov model and the starting vector P 0, which contains the probability of the patients of starting in each health state. The proportion of the initial cohort in each of the three states after one cycle ( P 1) can be calculated by multiplying P 1 by the matrix of transition probabilities, A. More generally the proportion of the initial cohort in each state after k cycles becomes P k= P k1*A, where P k -with dimension 1×sdisplay the proportion of the cohort contained in the defined states at cycle k. The structure for a Markov model will depend on the clinical application, the available data and how many simplifying assumptions are made. However, there are a number of essential steps to follow when constructing a Markov model: i) Specify the Markov states to reflect the relevant states of health and resources expenditure associated with the disease and treatment over time ii) Choose the cycle length to be used in the simulation, which must be a constant increment of time. The selected elapse of time should be short enough to consider the changes of clinical effects and resource use in 56 patients between the cycles. The time horizon for the analysis also should be chosen iii) A cost and utility should be assigned to each health state. In order to calculate discounted utility or cost, they should be divided by (1+r) k, where r is the discount rate corresponding to the cycle length and k is the cycle index84 iv) A set of transition probabilities must be specified. They indicate the chance of the individuals in the model to move from one health state to another. They can be defined as a function of time. For that purpose, a different matrix Ak for each cycle k should be defined to provide a transition probability linked to be health states and incorporate the time elapsed after an event. Introducing a statistical distribution (e.g., the exponential) or temporary and tunnel states can accomplish this purpose. The temporary states are used when a health situation has a short duration but has an important effect in costs or outcomes; the patients can only stay at the state for, at most, one cycle; their use enables the model users to assign state specific transition probabilities and adjust utilities and costs. The tunnel state, in which patients can only transit in a fixed sequence, is analogous (given the nature of life-threatening disease) to passing through a tunnel, and would be used when a temporary state would last more than one cycle85. The cohort simulation at the population level procedure consists on a hypothetical cohort of people who begin the process with some determined distribution among the states ( P 0). In the next cycle, the cohort is divided between the states according to the probability of transition, thus yielding a new distribution of the cohort between the states. This will continue in the subsequent cycles until the process has reached a cycle limit. The movement of the cohort through the health states during the simulated time produces estimations for the cumulative 57 utilities and costs. Table 3.1 illustrates the Markov trace for the first 2 cycles for the 3-state model used as an example (a numerical example can be followed in Annex VI). The simulation is run until the entire initial cohort resides in an absorbing state or until the upper limit of time that was considered clinically reasonable for the assessed health problem is reached. 58 Example: Markov model trace for the two first cycles is displayed below: Table 3.1. Two-cycle Markov trace for a 3-state Markov model with health states: Well, Sick and Dead. Utility scores for the health states are u W , u S , u D , respectively. Costs are defined analogously using the c as notation. P 0 =(1,0,0), i.e., P 0 =( p 0W , p 0S , p 0D ). Column 1 show the cycle number (k), columns 2 to 5 show the proportion of the cohort in each of the 3 health states at each cycle k (P k ), the last 2 columns show the utility and the cost contribution in each cycle # At cycle 0, the utility and the cost can be multiplied by 0.5 to take into account that some individuals transit in the middle of the cycle, which is known as the half-cycle correction. ¤ To obtain discounted expected utility (or cost) values the cycle utility (or cost) would be divided by its discount factor (1+r) k . Cycle (k) Well Sick Dead Cycle utility Cycle cost¤ 0 1 0 0 1* or 0.5* (pkW*uW+ pkS*uS+ pkS*uD)# 1* or 0.5* (pkW*cW+ pkS*cS+ pkS*cD)# 1 p0W * pWW+ p0S * pSW p0S * pSS+ p0W * pWS p0D * pDD+ p0W * pWD+ p0S * pSD (pkW*uW+ pkS*uS+ pkS*uD) (pkW*cW+ pkS*cS+ pkS*cD) 2 pk-1W * pWW+ p k-1S * pSW pk-1S * pSS+ pk-1W * pWS pk-1D * pDD+ pk1W * pWD+ p k-1S * pSD (pkW*uw+ pkS*us+ pkS*uD) (pkW*cw+ pkS*cs+ pkS*cD) 59 To draw a cohort simulation at the individual level we perform a first-order Monte Carlo simulation. The individual track is simulated, one at a time, through the tree of possible states. The first individual would start in the “Well” health state, based on P 0 , the next cycle visited will be determined using a random number drawn from a Uniform[0,1] and using the ordered cumulative probabilities for the cycle 1, i.e.: P 1 =(p 1W , p 1S , p 1D ) Assuming that p 1W >p 1S >p 1D , in case of equality the order can be decided at random. Then, the value in the [0, 1] obtained from a uniform distribution to allocate the individual in a health state for the cycle 1 is used as follows: If the uniform drawn value is in the [0, p 1D ] range the individual is going fall in the Dead health state (D). If it is in the [p 1D , p 1D +p 1S ] the health state is going to be Sick (S). Otherwhise (in the [p 1D +p 1S , 1]) the individual will reside, at least for the cycle 1, in the Well state. The simulation will be repeated for an individual until the dead state or the end of simulation time is reached; individual tracks would be performed up to the sample size wished for the cohort. The quality-adjusted life years (or cost) are calculated by taking the average of all the quality-adjusted life (or cost) spans in the cohort. The individual simulation has the advantage that conditional factors can be set up (e.g., conditional adherence) because the simulation is performed for individuals rather than for a full cohort. This approach offers plenty of flexibilities but often requires a very large number of simulations for accuracy of estimates. The standard error of the sample 60 mean can be estimated from a preliminary sample of, say n=1000. Since the standard error is quasi-proportional to the square root of n, the standard error with sample size N would be estimated to be roughly Sn*(n/N)1/2. The required sample size, N, depends on the magnitudes of the transition probabilities, the differences in utilities between states, and the effect sizes of interest. The HIV model by Freedberg et al. uses N=1,000,000 in order to obtain reliable estimates of cost-effectiveness ratios86. To seize parameter uncertainty for a cohort at population or individual level a probabilistic sensitivity analysis or a second-order Monte Carlo can be used. Both procedures require the input values to be extracted from a probability distribution. The results of the sensitivity analysis account for the variability in the input parameters. Even other distribution can fit the input parameters; cost can be drawn from a Gamma distribution, probabilities can be draw from a Beta or a Uniform distribution and utilities can be distributed as a Lognormal, Beta, or Uniform law. Several simulations are run using different input parameters. The analysis of the outputs obtained from these simulations provide a broad view of how much the variation in the inputs might affect the results and acts as a tool to check whether the assumptions made in the model definition are reasonable and do not influence the result. Both the cost-effectiveness acceptability curves (CEACs) and the scattered plot in the incremental cost-effectiveness plane are a good summary for the outputs of the sensitivity analysis and show the uncertainty of the model results. The CEACs depict the probability that each scenario is the most costeffective at any particular willingness to pay (or ceiling ratio) per unit of health gained. They are constructed by plotting the proportion of cohort simulations were each of the treatments assessed were cost-effective 61 for many ceiling ratios. It is noteworthy that the sum of the plotted proportions for every ceiling ratio value is 1. The other figure is obtained by displaying the pairs of cost and effectiveness values for every simulation over an incremental costeffectiveness plane. As it was described in Subsection 1.3.2. Usually, the scatter plot covers all four quadrants, indicating uncertainty about whether or not the intervention is cost-effective, and at what value it is cost-effective. The purpose of the CEAC is to summarise this uncertainty87. 3.2. Introduction to osteoporosis disease Thirty percent of the postmenopausal women suffer osteoporosis in Spain175. This diseases is characterized by low bone mass and structural deterioration of bone tissue, leading to bone fragility and an increased susceptibility to fractures, especially of the hip, spine and forearm, with vertebral fractures, although any bone can be affected88-90. Of all patients that developed a vertebral fracture, it is estimated that 20% will suffer a new vertebral fracture within a year91. Of all osteoporotic fractures, hip fractures are the most dangerous with an elevated mortality risk as well as a high hospital burden in Spain92. Osteoporosis has a negative impact on the health related quality of life (HRQoL) of the affected individual93. The increasing number of fractures due to osteoporosis in the past 20 years combined with the development of novel agents for the prevention or treatment of osteoporosis results in a health resources allocation problem94. Various treatments are approved for the prevention of osteoporotic fractures. Although they have been considered effective for the treatment of postmenopausal osteoporosis, some of them are not 62 appropriate for all women because of safety and/or tolerability issues95, 96. The selective estrogen receptor modulator (SERM) therapies, both raloxifene and bazedoxifene, had shown to reduce the risk of vertebral fractures in postmenopausal women97. Bazedoxifene has also associated with a favourable endometrial, ovarian, and breast safety profile in a 2year, phase 3 study of postmenopausal women at risk for osteoporosis98100. In Spain, approximately 2 million women101 were estimated to have osteoporosis in 2010. Treating this population is associated with a high socioeconomic burden and both clinical and economic implications should be taken into consideration to build a model to compare the treatment options to achieve higher long-term benefits of fractures risk reduction. Many models have been developed to study the socioeconomic impact of osteoporosis treatments for the Spanish National Health Service, as well as for patients102-105. Different tools are being used to estimate fracture risk which, at the same time, can vary significantly between countries105, these items can influence the results of any cost-effectiveness analysis. A recently published cost-effectiveness analysis comparing bazedoxifene with placebo used the FRAX® algorithm that provides fracture probabilities for specific populations105. Although FRAX® can be used to predict the probability of hip or other major osteoporotic fractures, the criteria should not be generalized to other countries having different fracture incidence rates and health care106. Therefore, when comparing the cost-effectiveness of bazedoxifene with raloxifene for Spanish osteoporotic women, it is important to take into account that the incidence of fractures is different for Southern European countries than countries in the Scandinavian region107, 108. The objective is to build a model to evaluate the cost-effectiveness of bazedoxifene and raloxifene for the prevention of vertebral and nonvertebral fractures among women diagnosed with osteoporosis, 63 accommodating the special characteristics of the disease in the Spanish setting. 3.3. Model to estimate the cost-effectiveness of bazedoxifene versus raloxifene Cost-Effectiveness analysis Our work sought to assess the cost-effectiveness of the available SERM treatments in terms of cost per QALY. The clinical evolution of the disease was based on the Osteoporosis Study109 and applied to the Spanish setting. The simulation model is implemented in Microsoft® Excel to calculate cost-effectiveness using an updated Markov model that has been used previously to estimate the cost-effectiveness of bazedoxifene incorporating the FRAX® algorithm using a European perspective105. The assessment was performed from the perspective of Spanish National Health Service and the time-horizon considered was 27 years, from 55 years old to 82 years old. The starting age was based on women recruited for bazedoxifene’s 3-year treatment clinical trial109 and 82 years old correspond to the life expectancy of a Spanish women110. QALYs gained was included as an effectiveness measure to allow us to compare the value of the interventions across different disease states. The incremental cost-effectiveness ratio (ICER), which is a measure of the added cost per QALY gained, is given as an output of this model. Decision analytic model The model evaluated the cost-efficacy of receiving bazedoxifene or raloxifene during this 27 year time. It was assumed that no patient discontinued treatment because of adverse effects. 64 Model specification The model simulated the transition of postmenopausal osteoporotic women through six defined health states (represented by ovals in figure 3.4) based on yearly transition probabilities. All patients began in the well-health state or no event state. In each cycle, a patient had a probability of sustaining a fracture, remaining healthy, or dying. After one year in any fracture state, the patient had a risk of sustaining a new fracture or dying. When a woman passes away, she would continue into the dead-health state for the rest of the simulation. After one year, the patient moved to the corresponding post-fracture state if no additional fracture occurred. The patient would automatically remain in the postfracture state (shown as a circular arrow) if she did not die or sustain a new fracture. Fractures could be vertebral or non-vertebral, consisting half of hip fractures and half of wrist fractures. After a non-vertebral fracture, it was possible to suffer a vertebral fracture or another nonvertebral fracture (Figure 3.4). Figure 3.4. Graphic representation of the simulation model. Ovals represent the health states and the arrows the possible transitions among them. 71 nowadays drugs cost (€286.52 and €171.86 per bazedoxifene and raloxifene treatment per patient year) and the results obtained show that the expected cost per patient was 1,292€ higher in the bazedoxifene cohort compared with the raloxifene cohort. The estimated QALYs gain was slightly higher in the bazedoxifene treatment branch than in the raloxifene one (0.02 QALYs). The PSA shows that bazedoxifene and raloxifene are almost equal in their probability of cost-effectiveness. The model was implemented using Microsoft® Excel Office 2007. It was built allowing the user to restore the original default parameters easily and to evaluate different possible scenarios, all input parameters were presented on one input worksheet and outputs displayed in several worksheets in a logical manner that summarizes the findings for the user, displaying tables and plots. The introductory worksheets describe the structure and the assumptions. These properties make this tool available and easy to use for the health care managers that should choose the best health care options. Other software options are available for implementing the Markov models for simmulation, such as R or Matlab. More specific programs designed with the aim of using decision trees and Markov models for decisions in an applied environment are also in the market –for instance, TreeAge©- but they are not simple enough to allow a basic user to change the input parameters to calculate the results for different scenarios. Conclusion This study investigated the cost-effectiveness of bazedoxifene compared with raloxifene in Spanish postmenopausal osteoporotic women and indicated that bazedoxifene was the dominant treatment strategy compared with raloxifene for the prevention of vertebral and nonvertebral fractures in postmenopausal osteoporotic women aged 55 to 82 72 years. The probabilistic sensitivity analysis that accounted for parameter uncertainty confirmed the deterministic results in a 52% of the realizations and did not create an evidence to select between treatments. The use of bazedoxifene supposes a small gain in terms of cost and QALYs and the decision between treatments should be reinforced with other clinical features not included in the model, such can be safety and tolerability113. Raloxifene was available later (November 2012), as a generic, for a lower cost than bazedoxifene. The cost-effectiveness analysis with the current prices showed that it can be a cost-effectiveness option when compared with bazedoxifene (Data not shown). 73 Summary Chapter 4. HIV TROPISM TESTING FOR MARAVIROC ALLOCATION. MARKOV MODEL IN N-STAGES Markov models were adapted to reflect that the risk of suffering an event can change over time. This analytical model was applied to elucidate which of 3 available co-receptors tests is cost-effective to determine patient’s suitability to benefit from the use of an antiretroviral treatment that includes maraviroc. All HIV strains require binding to CD4 plus at least one of the 2 co-receptors CCR5 or CXCR4 to enter human cells. Some HIV patients can use both co-receptors, and some individuals have a mixture of strains. Only patients with exclusively CCR5-tropic HIV are eligible to use the CCR5 antagonist maraviroc. The co-receptor assessment with 454 test or PS is nearly equal in effectiveness to Trofile-ES test but less expensive. Their Incremental Cost-effectiveness Ratios (ICER) were estimated to be 68,185 €/utility and 77,482 €/utility. There is not a dominating strategy; the expensive strategies also have a higher health outcome. The results of the PSA showed that the differences between tests are very small and we cannot claim the superiority of any of them. The choice will depend on the maximum that the health service is prepared to pay per additional unit of utility gained. 74 75 4. HIV TROPISM TESTING FOR MARAVIROC ALLOCATION. MARKOV MODEL IN N-STAGES The HIV/AIDS is a major health problem but antiretroviral (ART) regimens proved to be effective in decreasing HIV plasma viral load, improving CD4 cell counts, and have substantially altered the natural history of HIV infection. The introduction of new antiretroviral agents has broadened the number of active agents available for treatment of patients with infection due to HIV virus with certain particularities such as, its co-receptor type and/or the presence of drug resistance mutations. The new drugs in combination with new tools for the diagnosis have improved the success rate of therapy. In the case of the maraviroc, only patients with exclusively CCR5 HIV co-receptor (not CXCR4 either mixed-tropic virus) are considered eligible to use the CCR5 antagonist maraviroc. The objective of our work is to compare the cost-effectiveness of three different tests to determine HIV co-receptor usage (CCR5 and/or CXCR4) in order to select candidates for maraviroc. Markov models are used to simulate a cohort of patients at a population level and its path through different health states to calculate cost and health parameters. The resulting cohort will be used to assess the performance of diagnostic tests in HIV antiretroviral treatment allocation. This is an adaptation of the available methodology implemented to add flexibility to the Markov models. 4.1. HIV antiretroviral treatment and HIV co-receptor usage tests The HIV/AIDS is considered a pandemic, a disease outbreak that is not only present over a large area but is actively spreading114. Standard ART 76 therapy consists of the combination of at least three ART drugs to maximally suppress the HIV virus and stop the progression of HIV disease. Raltegravir, darunavir, maraviroc, and etravirine are new drugs frequently considered for use, particularly in ART experienced patients. Limited information exists regarding optimal combinations of these agents for the treatment. Selection of treatments combinations is often based on resistance testing results, prior treatment history, and any intolerance. Maraviroc was shown to be cost-effective, particularly in individuals with limited options for active antiretroviral therapy115. However, the role of maraviroc in this setting has been limited because of the high frequency of dual/mixed-tropic or CXCR4-tropic virus in patients with long-standing HIV infection and the necessity for expensive tropism assay testing116. Various strains of HIV use one of two co-receptors - CCR5 or CXCR4along with the CD4 receptor to enter human cells. Some HIV can use both co-receptors, and some individuals have a mix of strains (known as mixed-tropic virus). Only patients with exclusively CCR5-tropic HIV are considered eligible to use the CCR5 antagonist maraviroc, which blocks the virus from using this co-receptor. Patients susceptible to be treated by the drug are screened using a phenotypic viral tropism assay, the standard of care is the called Enhanced sensitivity Trofile test. As the MERIT-ES study demonstrated, accurate identification of patients with CCR5-tropic virus is an important predictor of treatment response117, 118. Recently, researchers have shown that a genotypic tropism test -the 454 sequencing-, or Population Sequencing test -PSmay perform well in predicting which patients will respond to maraviroc and other drugs in its class. Genotypic tests (which look at viral genetic sequences) are easier to perform than phenotypic tests (which look at how the virus behaves in a test tube), and therefore are usually less expensive (see characteristics in figure 4.1). 77 Figure 4.1. Description of the available tropism tests grouped by phenotypic and genotypic procedure. Their characteristics and their accuracy in virus detection are reported. Investigators retrospectively analyzed stored samples from a subset of 572 participants in the MOTIVATE-1 trial, which evaluated maraviroc versus placebo, combined with an optimized background regimen, in treatment-experienced patients119. They compared treatment response rates between patients identified as having CCR5 virus according to the genotypic test and the Trofile assay. Note that this study used the original Trofile test, not the enhanced sensitivity assay used in the MERIT-ES re-analysis. The genotypic test looked at the V3 loop of the HIV-1 gp120 protein, which plays a role in interactions between the viral envelope and host cell co-receptors. V3 genotype and standard Trofile were comparable in predicting antiviral responses to maraviroc in treatment experienced patients120. Despite 78 apparently poor sensitivity of standard genotyping for predicting nonCCR5 HIV relative to standard Trofile, these findings suggest the potential of genotyping as an accessible assay to select candidates for maraviroc. HIV V3 genotyping shows promise as a significantly faster and more cost-effective way to correctly identify patients who would benefit from CCR5 antagonists. Furthermore, the genotypic test is based on methods that are already widely used through the same labs that provide HIV drug resistance testing; this approach could become broadly available and be conducted at the same time as resistance testing to determine susceptibility to all drugs, including maraviroc. The model should be realistic and reflect all the variability that the test selection implies in the daily clinical practice, furthermore than the accuracy, the cost of the tests and the possibility to extend their use for all patients should be considered. We adapt the simulation based on Markov models allowing different phases of the evolution of the disease process characterized by different transitions probabilities matrices. The full program for cost-effectiveness and sensitivity analysis was implemented in R (see Annex IX). 4.2. Markov models in n-stages Markov models are a simulation tool frequently used in medical decision analysis. These models are especially appropriate when the disease of interest is characterized by the recurrence of particular events and when these are associated with a continuous risk over time81. The basic feature of the Markov model is that future events only depend on the current state that the patient is in, and not on prior events. A disease is characterized by using a finite number of health states and time is 79 handled as discrete periods of the same length. The implementation of the models is done assuming that the probability of travelling between health states is the same over time; flexibility can be added by introducing tunnel states and tolls, which make the model more complex. For some of the biological parameters assessed in cost-effectiveness studies, it is relevant to consider various phases on the evolution, which can be characterized for different probabilities of transition among the health stages defined in the structure of the Markov model. A 2-phase evolution process can be observed in several biological parameters such as the control of the HIV viral load, the recovery of CD4, CD8 or lymphocyte cells under active ART therapy or the serologic course of Hepatitis A-E virus infection under treatment121-125. The model adaptation performed in this thesis allows considering different matrices of transition probabilities to describe different phases of evolution for the disease course. The cohort simulation at a population level considers a hypothetical cohort of people were all members begin the process with some determined distribution among the states, usually designated according to the characteristics of our population of interest and/or the information found on the literature. In the next cycle, the cohort is divided among the states according to transition probabilities, which yields a new distribution of the cohort among the states. This continues in subsequent cycles until the process has reached the horizon time or the entire cohort reaches an absorbing state. For the model building several elements must be defined: i) A finite number of informative and realistic health states that can result from the evaluated therapies. The states should be mutually exclusive and collectively exhaustive 80 ii) The cycle length and the study horizon time. The time horizon should be equal to the sum of the lengths of the different phases considered iii) Costs and utilities assigned to every health state iv) The transition probabilities matrices must be specified. The number of transiton matrices is function of the number of therapeutic strategies assessed and they correspond to the number of Markov process to run. The performance of one Markov process was described in Section 3.1. When several phases of outcome evolution are considered, the probability of travelling between health states depend on the phase; this is reflected in the model by using different transition probability matrices. Example: A simulation by a Markov model in 2-phases is illustrated below, the 3 health states model introduced previously is used. Figure 4.2 represents the procedure of simulation to be run for every assessed therapeutic strategy. 87 Costs were reported from the third party health care payer perspective, acknowledging that Spain has universal health care coverage that includes tests and prescript antiretroviral ART drugs for this patient’s group (see disclosure in Table X.2). Considering the proportion of patients with adverse events (MERIT-ES study) the mean cost per patient/cycle were €3,161.13 for patients allocated to the Trofile-ES test, €3,067.38 for the patients allocated to the 454 test and €3,051.13 for the PS test group. The utilities related to the states of Undetectable, Detectable and Death were 0.83, 0.79 and 0, respectively. Cost and effectiveness annual discount rates were both set at 3%. 4.4. Results Analytical results The results were based on deterministic model calculations. The model estimated the average costs and utilities per patient of the lifetime horizon for the three groups of testing. The utility for patients screened with Trofile-ES test was similar to patients screened with 454 or PS test (10.67, 10.66, and 10.65 respectively; equivalently 3.557, 3.553 and 3.550 utilities per year). The utilities gained were less than 0.1 utilities. This indicates that all coreceptor tests have a very similar performance in guiding the therapeutic strategy. The expected cost per patient per 3 years of treatment was higher in patients tested with Trofile-ES test (€41,037; €13,679 per year) in comparison with patients in 454 test cohort (€39,821; €13,274 per year) 88 and the PS test (€39,609; €13,203 per year) with a difference of €1,216 and €1,428 (equivalent to €405 and €476 per year). This indicates that testing patients with the Trofile-ES test leads to more expensive treatment under the assumed conditions (Figure 4.6). 3.550 3.551 3.552 3.553 3.554 3.555 3.556 3.557 13200 13300 13400 13500 13600 Utilities per year Cost per year(€) Trofile 454 PS Figure 4.6. Cost efficacy plot for the three therapeutic strategies evaluated. The values for cost and utility for a year of simulation are displayed. Therefore, the testing with 454 test or PS is nearly equal in effectiveness as Trofile-ES test but less expensive. Their Incremental Cost-effectiveness Ratios (ICER) were estimated to be 68,185 €/utility and 77,482 €/utility. There is not a dominating strategy; the expensive strategies also have a higher economical cost. Model validation The results produced by a model are as reliable as the quality of the data used to generate the results. In the Markov cohort model, the estimated average effects (Utilities) and costs are the direct outcome measures, 89 but they are ultimately dependent on the accuracy of the HIV-tropism test. The model’s ability to translate test accuracy into patients with undetectable VLs is instrumental in the calculation of the ICER, which is the primary outcome measure that incorporates both costs and utilities. A penalization for the fact that a patient has to wait to know the HIVtropism was introduced in the model by decreasing the utility in a 5% for every week of turn-around test result (Scenario 8 in the Table 4.2). To assess the consequences of using concrete input parameters, the base case output was compared to the model output under a range of input parameters. A series of deterministic one and two-way sensitivity analyses were conducted to explore the impact on the ICERs of alternative assumptions for the values of key input parameters. The parameters and values tested generate a list of possible scenarios, which are described in the Table 4.2. 90 Table 4.2. Fifteen scenarios were created changing the input parameters to perform a sensitivity analysis. The Scenarios 1 to 8 are one-way analysis, and the following ones are two-way analysis Test Sensitivity Costs Utilities Comments Scenario 1 454, from 250€ to 150€ Scenario 2 454, from 250€ to 100€ Scenario 3 454, from 73% to 63% Scenario 4 454, from 73% to 83% Scenario 5 PS, from 60% to 50% Scenario 6 PS, from 60% to 70% Scenario 7 VL>50 from 0.79 to 0.69 Scenario 8 Trofile-ES, VL≤50: 0.622;VL>50: 0.592 454, VL≤50: 0.705;VL>50: 0.671 PS, VL≤50: 0.746;VL>50: 0.710 The utilities are reduced to penalize for the turnaround test. A week of waiting time reduces 5% the utility. Trofile-ES: 5 weeks; 454: 3 weeks; PS: 2 weeks. Scenario 9 454, from 250€ to 150€ Trofile-ES, VL≤50: 0.622;VL>50: 0.592 454, VL≤50: 0.705;VL>50: 0.671 PS, VL≤50: 0.746;VL>50: 0.710 Two-way sensitivity analysis: Scenario 1+Scenario 8 Scenario 10 454, from 250€ to 100€ Trofile-ES, VL≤50: 0.622;VL>50: 0.592 454, VL≤50: 0.705;VL>50: 0.671 PS, VL≤50: 0.746;VL>50: 0.710 Two-way sensitivity analysis: Scenario 2+Scenario 8 Scenario 11 454, from 73% to 63% Trofile-ES, VL≤50: 0.622;VL>50: 0.592 454, VL≤50: 0.705;VL>50: 0.671 PS, VL≤50: 0.746;VL>50: 0.710 Two-way sensitivity analysis: Scenario 3+Scenario 8 Scenario 12 454, from 73% to 83% Trofile-ES VL≤50: 0.622;VL>50: 0.592 454, VL≤50: 0.705;VL>50: 0.671 PS, VL≤50: 0.746;VL>50: 0.710 Two-way sensitivity analysis: Scenario 4+Scenario 8 Scenario 13 PS, from 60% to 50% Trofile-ES, VL≤50: 0.622;VL>50: 0.592 454, VL≤50: 0.705;VL>50: 0.671 PS, VL≤50: 0.746;VL>50: 0.710 Two-way sensitivity analysis: Scenario 5+Scenario 8 Scenario 14 PS, from 60% to 70% Trofile-ES, VL≤50: 0.622;VL>50: 0.592 454, VL≤50: 0.705;VL>50: 0.671 PS, VL≤50: 0.746;VL>50: 0.710 Two-way sensitivity analysis: Scenario 6+Scenario 8 Scenario 15 Trofile-ES, VL≤50: 0.622;VL>50: 0.517 454, VL≤50: 0.705;VL>50: 0.586 PS, VL≤50: 0.746;VL>50: 0.620 Two-way sensitivity analysis: Scenario 7+Scenario 8 91 The results of the series of sensitivity analysis performed to explore the impact of taking alternative assumptions for the input values on the results are displayed in the Table 4.3. Table 4.3. Results of the cost-effectiveness study, in terms of costeffectiveness ratios (ICERs), according to the various possible scenarios (units given in 2010 Euros) Analysis Trofile test Genotypic 454 test PS ICER Total Cost Total Utilities Total Cost Total Utilities Total Cost Total Utilities Trofile vs 454 Trofile vs PS 454 vs PS Base-case 41037 10.67 39821 10.66 39609 10.65 68185 77482 353767 Scenario 1 41037 10.67 39659 10.66 39609 10.65 77287 77482 83300 Scenario 2 41037 10.67 39578 10.66 39609 10.65 81837 77482 -51933 Scenario 3 41037 10.67 39821 10.66 39609 10.65 67504 77482 505381 Scenario 4 41037 10.67 39821 10.66 39609 10.65 68377 77482 326554 Scenario 5 41037 10.67 39821 10.66 39609 10.65 68185 75998 221104 Scenario 6 41037 10.67 39821 10.66 39609 10.65 68185 78418 558564 Scenario 7 41037 10.42 39821 10.36 39609 10.35 19209 22129 171177 Scenario 8 41037 8.42 39821 9.05 39609 9.58 -1916 -1231 -404 Scenario 9 41037 8.42 39659 9.05 39609 9.58 -2186 -1235 -95 Scenario 10 41037 8.42 39578 9.05 39609 9.58 -2314 -1235 59 Scenario 11 41037 8.42 39821 9.05 39609 9.58 -1916 -1231 -404 Scenario 12 41037 8.42 39821 9.05 39609 9.58 -1916 -1231 -404 Scenario 13 41037 8.42 39821 9.05 39609 9.58 -1916 -1231 -404 Scenario 14 41037 8.42 39821 9.05 39609 9.58 -1916 -1230 -404 Scenario 15 41037 7.81 39821 8.80 39609 9.31 -1230 -953 -416 Note: Values in black represent the changes with respect to the base-case analysis. A negative ICER means that the 2nd therapeutic strategy improves the utility and reduces the cost. A great variability can be observed from the results obtained for the different scenarios. As a summary, it can be said that, for a 46.7% (7/15) of the cases, the Trofile test showed to be more costly and more effective than the 454 test (ICER>0). In a 46.7% (7/15), the Trofile-ES was more costly and more effective than the PS test and for a 46.7% (7/15) the 454 test was more costly and more effective than the PS test. 92 In addition to the deterministic sensitivity analyses, a probabilistic sensitivity analysis was conducted. Probabilistic sensitivity analysis (PSA) The probabilistic simulation was performed by drawing each model parameter value from a specific probability distribution reflecting either patient’s individual characteristics or parameter uncertainty. The Beta distribution was used to generate the transition probabilities, and the utility values, the Gamma distribution was used for the costs. The distributions’parameters was computed by using the base-case value and its standard deviation assigned to be the 10% of the value, since these data were not available127. The utilities and cost of the 1,000 simulated trials per each of the three therapeutic strategies were computed. The cost-effectiveness ratios were plotted on the cost-effectiveness plane, and the cost-effectiveness acceptability curves were derived. It is required to define the ceiling Ratio and compute the net monetary benefit for each therapeutic strategy in order to plot the acceptability curve. The ceiling ratio indicates the amount of Euros that is worth to pay for the gain of one unit of health, 1 unit of utility in our case. The net monetary benefit for each therapeutic strategy was computed per every trial as the utility multiplied by the ceiling ratio minus the cost, and this value was used to assign a 1 to the therapeutic strategy that has the larger benefit, and a 0 to the other 2. A range of values of the ceiling ratio was used in order to plot the probability of each therapeutic strategy to show a larger benefit than the others. 93 The PSA showed results in which all treatments can be cost-effective since the density of the point estimates were spread in all quadrants of the cost-effectiveness plane (Figure 4.7). Regarding the mean of the incremental cost and utilities for the comparisons two by two it can be said that, in mean, the Trofile-ES test was dominated compared to the 454 test (Incremental cost=987.57, Incremental utility=-0.07); and also when it was compared with the PS test (Incremental cost=1,304.71, Incremental utility=-0.02). When comparing the 454 test and the PS, the first had a larger cost and a larger gain in health (Incremental cost=317.14, Incremental utility=0.04). -30000 -20000 -10000 0 10000 20000 30000 -5 0 5 Cost effectiveness plane (A vs B) Incremental utility gain Incremental cost(€) -30000 -20000 -10000 0 10000 20000 30000 -5 0 5 Cost effectiveness plane (B vs C) Incremental utility gain Incremental cost(€) -40000 -20000 0 20000 40000 -5 0 5 Cost effectiveness plane (A vs C) Incremental utility gain Incremental cost(€) Figure 4.7. Cost-effectiveness plane for the comparison between therapeutic strategies. A=Trofile-ES, B=454 test and C=PS. The figure 4.8 shows the probability that a treatment is the most effective of the three therapeutic strategies at a different threshold values for cost-effectiveness (ceiling ratio). For small willingness to pay quantities (under €10,000) the differences are small and the cheapest test seems preferable. For ceiling ratios from 10,000 onwards the difference between tests are very small and we cannot claim the superiority of any of them. 94 0 1020304050 0.00.20.40.60.81.0 Value of ceiling ratio (K €) Probability cost-effective A-Trofile B-454 C-PS Figure 4.8. Cost-effectiveness acceptability curve for the three therapeutic strategies by different values of the ceiling ratio. 4.5. Discussion and conclusion The limitations and the strengths of the model are detailed, the contextualization of these issues allow us to understand better the relevance and applicability of the conclusions. Study limitations There are several gaps in empirical data that need to be filled. Information is lacking, for instance, on the relation between test accuracy and treatment allocation. The hypothesis that a wrong treatment allocation drives to treatment failure is not right in 100% of the cases, since the ART treatment is composed by 3 or 4 antiretroviral drugs, even MRV is not active the other drugs can control the virus replication leading to an HIV-RNA undetectable viral load. This assumption avoided 95 adding the probability of failure as a parameter in the model. Treatment efficacy reduction due to poor compliance and aspects such as treatment switching were not included in the model. The sensitivity analysis tried to account for the variability generated by the previously described terms and other unknow ones, while extraassumptions about them were not added in the base-case analysis. A higher-order Markov model can include historical information on several patients’ health states in the probabilities of moving next into a health state or another 128. In addition, microsimulation models, which during the last 20 years have been increasingly applied in qualitative and quantitative analysis of public policies, can solve this issue. Their technique would let each patient start the simulation at a different risk of becoming undetectable and this risk changes over time129. The microsimulation requires a larger set of input parameters than the cohort simulation and, it usually, gives similar results. The efficacy data used in this analysis were taken from a NorthAmerican population. Compliance with treatment recommendations and consistency of refilling are also likely to differ between health-care systems and cultural settings. It was shown that insurance coverage for prescription drugs increases the probability of use 130. Thus, the availability of country-specific data when evaluating the costeffectiveness is of relevance. Strengths of the model The implemented model could accommodate 2-phases of evolution of the outcome studied, emulating what happens in “real” life. It was a useful tool to learn about the cost-effectiveness of the three assessed tests to determine the HIV co-receptor without the need of doing a prospective clinical study. 96 Discussion The use of HAART had reduced the viral replication and reconstituted immunity, leading to longer periods of symptom-free disease and survival after AIDS diagnosis, and to changes in the natural history of HIV-associated illnesses. This encompasses an increase in the number of individuals that require treatment, taking into account the limited resources that can be spent on health services, the economic assessment for new antiretroviral medications are of interest for the health decisionmakers to optimize the use of health care resources. Cost-effectiveness analysis aim to provide information on the value of a new co-receptor test compared to the standard intervention. Costeffectiveness does not necessarily mean cost-saving; the total cost of a therapeutic strategy can be higher, but still considered good value for money if it enhances significantly the health outcome relative to the current standard. The model performed is an attempt to simulate a real world process using input data describing physical characteristics of the system, a set of algorithms to transform input data to output parameters of interest and simplifying assumptions to limit the scope of the model. The accuracy of the output measures depends on the quality of the input parameters and the structure of the model. The input parameters are estimations that have an implicit variability, which was not considered in the modelling process, but sensitivity analysis measured how this uncertainty can affect the results. The time horizon was settled to 3 years, longer simulation times can be unrealistic for the following reasons:  the patient’s characteristics change over the time,  therapeutic strategies and SOC can evolve,  and new variables of decision to allocate test and treatment can be identified as relevant. 103 rates of adoption of new therapies and sensitivity analysis. The Spanish national regulatory agency does not have any guidelines or recommendations. A literature review published in 2005 by Mauskofp et al.144 indicate that the number of studies in peer-reviewed journals is limited and varied greatly in the methods they used. Estimations of the financial impact for different timeframes and/or target patients were described: i) for a timeframe of 1 to 3 years ii) for lifetime costs for a specific cohort iii) for a set of representative individuals being started on competing treatments. A more limited number of published studies attempt to explicitly estimate the financial and health-care service impact of a new therapeutic strategy for a well-defined national or health plan population. Instead of publication, budget impact analyses are more frequently presented directly to decision makers as interactive computer programs designed to calculate the financial impact for specific health plans. There is also the ongoing debate about whether a BIAs should be totally or partially publicly available for review. BIA methods Whereas an economic analysis addresses the additional health benefit gained from the resources invested in health, the BIA addresses the affordability of a new therapeutic strategy. Several factors, which are not generally needed for cost-effectiveness analysis, should be part of a comprehensive budget impact analysis including the size of the treated population, second-order costs, market diffusion rates for the new drug, and off-label use137. See in Figure 5.1 how the information of BIA and the 104 incremental cost-effectiveness ratio (ICER) as a summary of a costeffectiveness analysis (CEA) can be obtained. Figure 5.1. Schematic representations of the budget impact and ICER computation. Both approaches compare the current environment with a new one –which can include a new therapeutic strategy-. BIA reflects resources use and the ICER considers both costs and health outcome. Adapted from Brosa et al.145. MD stands for medical doctor. The quantification of the affected population out of the total and the definition of the target population that can be benefited from the use of the health care intervention assessed is relevant, these numbers are going to frame the cost analysis. For instance, if the interest is to assess the congenital toxoplasmosis, the population of interest should be reduced to pregnant women. The epidemiologic information, the demographic data, and some risk factors should be taken into account to form the right frame for the study. 105 The resources used to treat a particular health condition should be listed as detailed as possible, but the cost assessment will be quantified for some selected items (or all) according to the cost component (or components) and the perspective of interest. Costs in health care can be subdivided into three components: Direct costs, indirect costs, and intangible costs. The items included in each component are described as follows:  The direct costs reflect the amount of money spent on medical products and/or services as a direct result of an illness  The indirect costs are the lost potential productivity resulting from illness-related absences or impaired performance at the workplace and at patient’s normal life activities  The intangible costs include humanistic measures of changes in health status such as health related quality of life, joy, and satisfaction or the cost of worries, pain, and suffering. These items are often included in the health outcome quantification by means of a health related quality of life or utility score. Further definitions about cost are given in Annex XI. The direct costs are relatively easy to measure. The indirect costs are considered costs from the perspective of society as a whole. Many of these are difficult to measure, and there is some controversy over which ones to include in the list and how to measure them. For instance, the UK National Institute for Clinical Excellence, NICE, adopts a limited societal perspective in its evaluations and considers the direct costs falling on the UK National Health Services, and those indirect costs funded by the state such as unemployment and sickness benefits146. Two perspectives can defined in terms of the view point of the analysis: If the health care payer point of view is selected, the study only accounts for the direct costs of health care for this specific payer and if the study 106 is conducted from the perspective of the society as a whole, direct, and indirect cost are both important. In general, the societal perspective is considered the most appropriate, but a health care manager with a limited budget might be tempted to ignore the societal view and consider only the costs that affect his own budget. An example of this situation is a study of migraine performed under the health service perspective only suggested that sumatriptan in migraine (an expensive drug in comparison with a cheaper treatment as the standard of care) was highly undesirable, but a study taking a societal perspective came to the opposite conclusion147. This example drives us to note that the comparison between 2 or more treatment scenarios is possible by repeating the cost calculations under the current environment and under the environment where the new treatment strategy is used, which is displayed in the last column of the figure 5.1, labelled as “New environment”. Two important concepts are the opportunity cost and the marginal cost. The opportunity cost is defined as the benefit foregone when selecting one alternative intervention (treatment A) over the next best alternative (treatment B). The opportunity cost of investing in a healthcare intervention (treatment A) is best measured by the health benefits that could have been achieved if the same amount of money had been spent on the next best alternative intervention (treatment B). The marginal cost is the resource cost associated with the use of treatment A in spite of treatment B, being an indicator of the amount of additional resources that must be expended or can saved (see detailed explanations in Annex XI). The perspective and items included in the analysis should be defined clearly and the data collection needs to be reflecting values as updated and accurate as possible. The validity of the results depends on the quality of this data. 107 The first models on value of health were developed in the insurance industry to assess and to characterize both the risk of suffering a disease, and the population health-care cost. The data used in the modelling was the recorded by the health insurance claims. During the last years, the study and prediction of the cost became an important subject of research to guide the health resources allocation; some of the authors have applied different statistical techniques to do the estimations and prediction148. Some published examples of budget impact analyses are described in the review by Mauskopf et al144. As time goes by, more data derived from administrative databases and/or expert opinion is available. In order to predict the financial impact the most common approaches are  to estimate the direct costs of the treatment of a health condition by assuming a linear behaviour in the cost for the near-term years and updating the proportion on the target population.  Alternatively to impose a tax of increase of target population into a Markov model incorporating clinical and epidemiological data to simulate the target population throughout the timeframe analysis. This method is suitable for diseases with rapid evolution.  Regression models are also used to characterize the cost as dependent variable in function of a set of independent variables selected by their clinical and economical relevance. The use of the coefficient estimated by the model can be used to predict the mean cost for next years.  Data mining tools were the newest and accurate approach that Bertsimas et al. applied to provide predictions of future healthcare costs. With the use of decision trees and clustering algorithms along with claims data from over 1,000,000 insured individuals over three years discovered that the pattern of past cost data is a strong predictor of future costs and medical 108 information provides an accurate prediction of medical costs particularly on high-risk members148. 5.2. The paroxysmal nocturnal hemoglobinuria Paroxysmal nocturnal hemoglobinuria (PNH) is a rare, genetically acquired blood disorder characterised by chronic intravascular hemolysis (destruction of red blood cells). PNH is clinically defined by the lack of the complement inhibitory protein CD59 on the surface of red blood cells. CD59 normally blocks the formation of the terminal complement complex on the surface of the red blood cell, which prevents hemolysis. The list of signs and symptoms of PNH include hemoglobinuria (presence of blood in the urine), anaemia, fatigue, difficulty swallowing, abdominal pain, erectile dysfunction in men and thrombosis149-151. PNH is most common among men in their 20s, but it occurs in both genders and at any age, causing high morbidity and mortality. It is estimated that rare diseases -those whose frequency is under 5 cases / 10,000 peopleaffect about 6% of the European population. The PNH is considered a rare disease which affects an annual rate of 1-2 cases per million152. Hill et al.157 estimated that the annual incidence is 1.3 cases per million, and the prevalence is of 15.9 cases per million of inhabitants. Men and women are affected equally and PNH may occur at any age but it frequently is found among young adults with a median age at the time of diagnosis of around 42 years (range, 16-75 years). This disease process is insidious and has a chronic course, with a median survival of about 10.3 years after the diagnoses153. PNH is the only hemolytic anaemia caused by an acquired intrinsic defect in the cell membrane. In the field of anaemia, 1% of couples are at risk of having a newborn with a severe syndrome of haemoglobin such are 109 sickle-cell disease or thalassemia. More than 330,000 children are born worldwide each year affected by one of these disorders. In Spain, the average risk of having a newborn with a rare or unusual anaemia has increased due to African immigration154. Treatment of PNH consists on supportive care measures including corticosteroids, androgen hormones, iron and folate supplementation, sometimes transfusions (generally reserved for crises) and allogenic stem cell transplantation which have been successful in a small number of cases until the development of a new drug known as eculizumab. This is a recombinant humanized monoclonal antibody that works by binding to complement protein C5, inhibiting its enzymatic cleavage, blocking the formation of the terminal complement complex, and thus preventing red cell lysis*. In 2007, the drug (eculizumab, Soliris™; Alexion Pharmaceuticals, Inc., Cheshire, CT), received approval as an orphan drug by the U.S. Food and Drug Administration (FDA) and by the European Medicines Agency (EMEA) for the treatment of patients with PNH to reduce hemolysis155. The FDA approval was based mainly on a randomized, double-blind, placebo-controlled, clinical trial in 87 RBC transfusion-dependent adult PNH patients, with supportive evidence from two observational studies:  A phase II pilot study involving 11 PNH transfusion-dependent patients, and  a 52-week, open-label, non-placebo-controlled, single-arm study in 96 PNH patients156. Two clinical trials to study the efficacy of eculizumab were published; Sheperd165 and Triumph 162, both has shown that eculizumab reduces intravascular hemolysis after the first week of treatment. The control of the hemolysis reports anaemia diminution and consequently the required * Lysis is defined as destruction or decomposition, as of a cell or other substance, under influence of a specific agent. 110 blood transfusions were reduced by 51%. Fatigue was also reduced after a few weeks of treatment, and the release was maintained up to the end of the study. Also, 3 phase II * studies concluded that eculizumab treatment leads to less blood transfusions need and a reduction on risk of thrombosis events157, 158. The aim of this study was to assess the budgetary impact in Spain of using eculizumab as a newly approved pharmacotherapy for PNH compared to the standard of care. Two perspectives were used: the health care system and the societal care system considering the cost of the comorbidities and the patient’s inability to work and perform a “normal life”. 5.3. Model for assessing the PNH treatment Model A budgetary impact model using Microsoft Excel Office 2007 following the international recommendations has been elaborated to estimate healthcare costs of the extended use of eculizumab as a treatment for PNH. The model was implemented allowing the user to restore the original default parameters easily and to evaluate different possible combinations. All input parameters were presented on one input worksheet and outputs were displayed in several worksheets that summarize the findings for the user, displaying tables and plots. The introductory worksheets describe the structure, assumptions, and use of * There are 5 phases for describing the clinical trial of a drug based on the study's characteristics. Phase 2 studies gather preliminary data on effectiveness (whether the drug works in people who have a certain disease or condition). For example, participants receiving the drug may be compared with similar participants receiving a different treatment, usually an inactive substance (called a placebo) or a different drug. Safety continues to be evaluated, and short-term adverse events are studied. [Reference: http://www.clinicaltrials.gov/ct2/info/glossary#P] 111 the model furthermore than all sources and assumptions associated with the input (see Annex XII and Annex XIII). Published clinical guidelines, clinical literature and expert opinion were employed to describe the required treatment and progress-over-time of patients affected by PNH. The analytical model was built as a flexible tool to predict the potential financial impact when changing some input parameters. Population The eligible population was obtained from estimates of the number of Spanish citizens affected by PNH. Treatment options Two therapeutic strategies are applied in the Spanish region: the standard of care consisting in blood transfusion, anticoagulant treatments, hematopoietic stem cell transplantation and/or bone marrow transplantation and the new treatment with eculizumab, which was the only treatment approved for the PNH159, 160. The standard of care has a limited and variable efficacy a non ignorable amount of adverse events that require continuous concomitant treatments complicate tackling the disease159, 161. Two scenarios were compared in the analysis: patients treated with blood transfusion versus the eculizumab treatment. These are the two common therapeutic strategies used in practice. An expected 15%, 35%, 55%, 70%, 85% and 100% of patients receiving eculizumab treatment were considered in the model in the first, second third, fourth and fifth year, respectively. It was assumed that patients with PNH are going to switch progressively from the standard of care treatment to eculizumab. 112 Time horizon, perspective, and discounting The analysis was performed with a 5-year time projection from the public payer and societal perspective. The costs were in Euros (EUR, 2010) and were reported with and without discounting by a 5% annual rate135, 136. The undiscounted values are given because in the good practices of the International Society for Pharmacoeconomics and Outcomes Research (ISPOR) reports is stated that budget impact analyses present financial streams over time and it is not necessary to discount the costs. However, some national guidelines published for BIA performances advise to apply a discount 6% annual rate, instead of 5% as we did. Resources and costs The model has been developed considering the costs associated to the standard of care and eculizumab treatments using the evolution of the disease and treatment efficacy and adverse event described in the Hillmen et al.162-164, Brodsky et al.165 and Kelly et al.166 research papers. The base-case analysis was defined as the eculizumab treatment and the alternative scenario was defined as the standard of care, mainly based in blood transfusions. The following resources and treatment components were identified and quantified as direct costs: drug costs, dispensing and administrating costs, the cost of treating the adverse and PNH related events. In the indirect cost setting, patients’ traveling expenses to attend to the health center and the loss of production due to PNH were included. Eculizumab should be administrated in doses of 600mg per week during the first 4 weeks of treatment, 900mg in the 5th week and after the 6th week 1 dose of 900mg every 2 weeks166. The cost of eculizumab for the first year is 342,650€ and for the 2nd and consecutive ones is of 320,400€ per patient year (PPY). The mean cost PPY is 315,479€. 119 Table 5.3. Budget impact analysis of the use of eculizumab versus the standard of care for 5 years % Patients Treated Now 1st year 2nd year 3rd year 4th year 5th year Eculizumab 15% 35% 55% 70% 85% 100% Standard of care 85% 65% 45% 30% 15% 0% Total annual direct and indirect costs Now 1st year 2nd year 3rd year 4th year 5th year Eculizumab € 4,865,739 € 11,466,926 € 18,199,649 € 23,394,822 € 28,692,078 € 34,092,940 Standard of care € 34,296,158 € 26,488,739 € 18,521,741 € 12,471,306 € 6,298,009 € 0 Discounted total annual costs Now 1st year 2nd year 3rd year 4th year 5th year Eculizumab € 4,865,739 € 10,893,579 € 16,425,183 € 20,058,135 € 23,369,877 € 26,380,467 Standard of care € 34,296,158 € 25,164,302 € 16,715,871 € 10,692,586 € 5,129,768 € 0  Standard of care Eculizumab Figure 5.4. Budget impact of the use of eculizumab versus the standard of care for 5 years while the percentage of patients treated with eculizumab increases over time. Treating patients suffering from PNH with eculizumab instead of the Standard of care resulted in a reduction in total societal costs of €79,102 per patient year, even the direct costs are larger when treating patients with eculizumab. Now 1st year 2nd year 3rd year 4th year 5th year 120 5.5. Discussion and conclusion Discussion The present work is an update of the current situation in terms of a number of patients and costs of the PNH. The goal is to provide details of the resource consumption and more accurate estimates of the budgetary impact, based on data following the complete clinical situation, including adverse events and events related to the studied disease. There are several limitations to be considered in this model, including the very limited number of published reports regarding resource consumption in PNH treatment. Prospective studies conducted under standard clinical practice, designed to collect resources and costs data associated to PNH would be desirable and could provide reliable information to be used in further economic evaluations168. Potential improvements in health related quality of life and benefits resulting from better health at a social level were included in the present model, but future works could be even more useful if a health score would be added, such can be the utility or a burden of disease score. An increase in the number of patients under PNH seems to be realistic; this was reflected in the budget impact analysis for 5 years using the 1% of patients growing tax. Conclusion The introduction of new (and expensive) pharmaceutical products is one of the major challenges for health systems169. The use of eculizumab for treating the PNH would imply an incremental yearly cost of €300,650 per patient compared to standard of care, but would provide larger societal benefits and an improvement in health related quality of life of the PNH affected patients leading to overall savings. These results would help the Regional, and National Authorities to perform a better allocation of 121 available resources, decisions in incorporating the new treatments to the guidelines and into the new standard of care can be taken with objective information. 122 123 Summary Chapter 6. DALYS IN POSTMENOPAUSAL WOMEN WITH OSTEOPOROSIS. HEALTH BENEFITS QUANTIFICATION Most of the published clinical studies are focused on measuring health in terms of efficacy and/or safety. Sometimes health and well-being quantification is not a direct measurement. The calculation of the burden of disease for osteoporotic women who may suffer from fractures done at an individual level was presented in terms of disability adjusted life years (DALYs). It was quantified that Mean (SD) overall undiscounted DALYs lost per woman were 6.1 (4.3), with a significantly higher loss in women with severe osteoporosis with prior bone fracture (BF); 7.8 (4.9) compared with osteoporotic women (5.8 (4.2)) or postmenopausal women with a BMD >-2.5 T-score after receiving a drug-based therapy (6.2 (4.3)). Factors explaining the variation in the levels of health were the alcohol consumption, having rheumatoid arthritis, previous osteoporotic bone fracture, family history of osteoporosis, using corticosteroids and a lower BD revealed to be linked to a larger DALYs lost. Few studies of burden of diseases are available, and even less for Spanish population and performed using individual characteristics. The identification of risk factors can improve the clinical practice by guiding the concerns that should be considered in the osteoporosis prevention. 124 125 6. DALYS IN POSTMENOPAUSAL WOMEN WITH OSTEOPOROSIS. HEALTH BENEFITS QUANTIFICATION * Different approaches are available for measuring the health benefit of any therapeutic strategy or any intervention performed. Historically, mortality rates have been used to describe health status across communities. These measures do not fully account for the burden of premature mortality, an important indicator of a population health. In fact, since most deaths occur among persons in older age groups, mortality rates are dominated by the underlying disease processes of the elderly170. Premature mortality entails estimating the average time a person would have lived if he or she had not died prematurely. This estimation inherently incorporates age and death, rather than merely the occurrence of death itself171. Over the last decades the need of measure the health outcome has increased. The introduction of an objective summary to quantify the health status is needed to better explain health across populations and to compare different treatments and/or individual groups (see further definitions in Annex II). When morbidity is taken into account, the two dominating summary measures are the Quality Adjusted Life Years (QALYs), and the Disability Adjusted Life Years (DALYs). QALYs and DALYs represent an implicit trade-off between quantity for quality of well-being. In QALYs, premature death is combined with morbidity by attaching a weight to each health state such that value 0 represents death, while value 1 represents full health. The number of QALYs for a health profile is found by multiplying the health related quality of life weight (HRQoL) of the health state, with the duration of the health state. Like the QALY, the DALY measure facilitates comparisons of all types of health outcomes by * TheDALYcalculationsandsomeofthestatisticalanalysispresentedinthischapterhavebeen performedalongwithLisetteKaskens 126 attaching disease weights were value 0 represents full health and value 1 represents death. Note that these disease weights are the opposite of the HRQoL weights in the QALY. A DALY can therefore be seen as an inverse QALY172. The goal of this chapter is to assess the burden of disease of the osteoporosis in postmenopausal women. DALYs are computed using individual information. Section 6.1 describes how the DALYs can be calculated. Section 6.2 contains the details of the calculation of the DALYs for the postmenopausal women with osteoporosis, the details of the available data and the statistical analysis. The description of the health quantification by means of DALYs and the factors associated to a higher burden of disease are given in section 6.3. The discussion and conclusions are displayed in the last section. 6.1. DALYs calculation The description of the characteristics of the disability-adjusted life year (DALY) for individual data is presented in the followoing. DALYs can be calculated according to Fox-Rushby et al.173. Previous to the DALYs computation we must state some definitions: K is the standard age-weighting modulation factor; C is a constant; r is the discount rate, usually r=0 or 0.03; a is the age of death or the age of onset of disability, for the calculation of Years of life lost (YLLs) and Years of life lived with disability (YLDs), repectively; β is the parameter from the age weighting function 127 L is the standard expectation of life at age a or the duration of the disability for the calculation of YLLs and YLDs, repectively; D is the disability weight. Calculations done using the individual data of the subjects in the sample: 1) Utility (U) and disutility (D) values: UD   1 The disutility value can be used in the following formulas as the disability weight. 2) Years of life lost (YLLs): The term YLLs is calculated using the following formula:           rL araLr ra e r K areaLre r KCe KrYLLs         1 1 11,, 2    3) Years of life lived with disability (YLD) is equivalent to the YLLs (adapting the definition of a and L) multiplied by the disability weight, in the formula noted as D:      ,,,, KrYLLDKrYLDs   4) Life expectancy (LE) at a particular age is   aLEaLE   5) And years of life lived with disability at age a is the product of disability weight and duration of disability at age a, i.e.    DadisabilitywithlivedLifeaYLD *. 6) DALY equals to the sum of YLLs and YLDs. YLDsYLLsDALYs   . 128 For the calculations at steps 2 and 3: The values introduced for r, K, and β were the ones recommended by Murray et al.196: r=0.03, K=1 and β=0.04. C is a constant with value 0.1658. These values constitute the base case analysis. The YLLs and YLDs can be computed without a discount rate (r=0) and without age weighting (K=0). The life expectancy, LE, was extracted from the Spanish national statistics database174. The information of the individuals recorded in the data set was used in the YLLs and YLDs calculations. The onset of the disease, life expectancy at a particular age and the disability weight were included in the formulas in order to obtain a particular value for each individual. It is noteworthy that 4 values for the DALY can be computed for every individual in the data set, depending on the discounting and age weighting combinations: Discounted, weighted by age; Discounted, non-weighted by age; Non discounted, weighted by age and Non discounted, nonweighted by age. The DALYs calculation using individual information is illustrated in the case of a sample of postmenopausal women. The reported DALYs are with and without a discount rate and without age weighting. Once calculated, the DALY values are analyzed as a dependent variable with the goals to estimate and describe the burden of the disease. 6.2. DALYs calculation for postmenopausal women with osteoporosis using individual information The osteoporosis disease and its clinical importance was described in section 3.2. 231 CROSSTABS /TABLES=tabacon alcohn gramosn BY DALYs_available /FORMAT= AVALUE TABLES /STATISTIC=CHISQ /CELLS= COUNT ROW COLUMN /COUNT ROUND CELL . **Background* CROSSTABS /TABLES=dmn h_artn art_reumn anor_nern hiperparan hipertiron hepat_cron sind_malabn BY DALYs_available /FORMAT= AVALUE TABLES /STATISTIC=CHISQ /CELLS= COUNT ROW COLUMN /COUNT ROUND CELL . ***Osteoporosis data*. T-TEST GROUPS = DALYs_available(0 1) /MISSING = ANALYSIS /VARIABLES = edad_diag dxan Nfrfo Frmosteo tiemdiagn /CRITERIA = CI(.95) . CROSSTABS /TABLES=imc_20_cat BY DALYs_available /FORMAT= AVALUE TABLES /STATISTIC=CHISQ /CELLS= COUNT ROW COLUMN /COUNT ROUND CELL . ********** Logistic regression***. **All variables***. LOGISTIC REGRESSION VARIABLES DALYs_available /METHOD = ENTER edadn sit_labn niv_estn habitatn imcn cign tabacon alcohn gramosn dmn h_artn art_reumn anor_nern hiperparan hipertiron hepat_cron sind_malabn edad_diag dxan Nfrfo Frmosteo tiemdiagn /CONTRAST (sit_labn)=Indicator(1) /CONTRAST (niv_estn)=Indicator(1) /CONTRAST (habitatn)=Indicator(1) /CONTRAST (tabacon)=Indicator(1) /CONTRAST (alcohn)=Indicator(1) /CONTRAST (gramosn)=Indicator(1) /CONTRAST (dmn)=Indicator(1) 232 /CONTRAST (h_artn)=Indicator(1) /CONTRAST (art_reumn)=Indicator(1) /CONTRAST (anor_nern)=Indicator(1) /CONTRAST (hiperparan)=Indicator(1) /CONTRAST (hipertiron)=Indicator(1) /CONTRAST (hepat_cron)=Indicator(1) /CONTRAST (sind_malabn)=Indicator(1) /PRINT = GOODFIT CI(95) /CRITERIA = PIN(.05) POUT(.10) ITERATE(20) CUT(.5) . **Assess and avoid multicolinearity**. LOGISTIC REGRESSION VARIABLES DALYs_available /METHOD = ENTER edadn sit_labn niv_estn habitatn imcn tabacon gramosn dmn h_artn art_reumn anor_nern hiperparan hipertiron hepat_cron sind_malabn edad_diag dxan Nfrfo Frmosteo tiemdiagn /CONTRAST (sit_labn)=Indicator(1) /CONTRAST (niv_estn)=Indicator(1) /CONTRAST (habitatn)=Indicator(1) /CONTRAST (tabacon)=Indicator(1) /CONTRAST (gramosn)=Indicator(1) /CONTRAST (dmn)=Indicator(1) /CONTRAST (h_artn)=Indicator(1) /CONTRAST (art_reumn)=Indicator(1) /CONTRAST (anor_nern)=Indicator(1) /CONTRAST (hiperparan)=Indicator(1) /CONTRAST (hipertiron)=Indicator(1) /CONTRAST (hepat_cron)=Indicator(1) /CONTRAST (sind_malabn)=Indicator(1) /PRINT = GOODFIT CI(95) /CRITERIA = PIN(.05) POUT(.10) ITERATE(20) CUT(.5) . **Final model include all the significative ones, removing from the multivariate model the non significative ones (one by one) **Avoiding multicolinearity and without including Dexa**. *** Selected model to be reported in the reviewer answer ****. LOGISTIC REGRESSION VARIABLES DALYs_available /METHOD = ENTER imc_20_cat Nfrfo Frmosteo /CONTRAST (imc_20_cat)=Indicator(1) /PRINT = GOODFIT CI(95) /CRITERIA = PIN(.05) POUT(.10) ITERATE(20) CUT(.5) . 233 Bibliography and references 1 Moore GE. Cramming More Components onto Integrated Circuits. Electronics. April 19, 1965: 114–117. 2 Krueger A. 6 Ways Technology Is Improving Healthcare. Business Insider, 20 December 2010. Available from: http://www.businessinsider.com/6-waystechnology-is-improving-healthcare-2010-12?op=1 Accessed 2 October 2013. 3 DNA Sequencing Costs. Available from: http://www.genome.gov/sequencingcosts/ Accessed 10 September 2013. 4 Moghimi SM, Hunter AC, Murray JC. Nanomedicine: current status and future prospects. FASEB J. 2005 Mar;19(3):311-30. Available from: http://www.fasebj.org/content/19/3/311.full#content-block Accessed 2 October 2013. 5 WHO. Health topics. Available from: http://www.who.int/topics/noncommunicable_diseases/en/ Accessed 3 June 2014. 6 Busse R, Miriam Blümel M, Scheller-Kreinsen D, Zentner A.Tackling chronic disease in Europe. Strategies, interventions and challenges. Brussels: European Observatory on Health Systems and Policies. Available from: http://www.euro.who.int/__data/assets/pdf_file/0008/96632/E93736.pdf Accessed 3 June 2014. 7 Singh D. How can chronic disease management programmes operate across care settings and providers? World Health Organixation 2008 and World Health organization, on behalf of the European observatory on Health Systems and Policies, 2008. Available from: http://www.euro.who.int/__data/assets/pdf_file/0009/75474/E93416.pdf?ua=1 Accessed 3 June 2014. 8 The crisis, hospitals and healthcare. Available from: http://www.hope.be/05eventsandpublications/docpublications/86_crisis/86_HOP E-The_Crisis_Hospitals_Healthcare_April_2011.pdf Accessed 2 October 2013. 9 Fiscal monitor: taxing times (Appendix 1. Recent Developments in Public Health Spending and Outlook for the Future). International Monetary Fund. October 2013. Available from: http://www.imf.org/external/pubs/ft/fm/2013/02/pdf/fmappx.pdf Accessed 28 January 2014. 10 World Health Organization Health systems’ responses to the economic crisis in Europe. Fact Sheet 12 December 2013. Available from: http://www.euro.who.int/__data/assets/pdf_file/0003/236676/Health-systemsresponses-to-the-economic-crisis-in-Europe.pdf Accessed 28 January 2014. 11 Saha S, Hoerger TJ, Pignone MP, Teutsch SM, Helfand M, Mandelblatt JS; Cost Work Group, Third US Preventive Services Task Force.The art and science of incorporating cost effectiveness into evidence-based recommendations for clinical preventive services. Am J Prev Med, 2001. 20(3 Suppl): p. 36-43. 12 Frequently used terminology. ISPOR. Available from: http://www.ispor.org/Terminology/ Accessed 2 October 2013. 13 INE. La esperanza de vida al nacimiento aumenta en más de dos años desde 1995 y se sitúa en 80.23 años. 29 October 2007. Available from: http://www.ine.es/prensa/np472.pdf Accessed 2 October 2013. 234 14 Guijarro M, Peláez O. La longevidad globalizada: un análisis de la esperanza de vida en España (1900-2050). Scripta nova, 1 de marzo de 2008. XII (260). Available from: http://www.ub.edu/geocrit/sn/sn-260.htm Accessed 2 October 2013. 15 Statistical Office of the European Communities. Eurostat. Available from: http://epp.eurostat.ec.europa.eu Accessed 12 August 2013. 16 Instituto Nacional de Estadística (INE). Notas de prensa 14 May 2013. Encuesta Nacional de Salud 2011 – 2012. Available from: http://www.ine.es/prensa/np770.pdf Accessed 3 June 2013. 17 Arjona Mateos CR, Criado Velasco J, Sanchez Solis L. Enfermedades crónicas y consumo de fármacos en mayores de 65 años. MEDICINA GENERAL 2002; 47: 684-695. Available from: http://www.mgyf.org/medicinageneral/revista_47/pdf/684-695.pdf Accessed 3 June 2013. 18 Sánchez Sánchez C. Impacto sociosanitario de las enfermedades neurológicas en España. Fundación Española de Enfermedades Neurológicas (FEEN) 2006. Available from: http://www.fundaciondelcerebro.es/docs/imp_sociosanitario_enf_neuro_es.pdf; Accessed 29 May 2013. 19 Notas de prensa. Sociedad española de cardiología. Available from: http://www.secardiologia.es/actualidad/notas-de-prensa/2876-valenciaacogera-el-congreso-de-las-enfermedades-cardiovasculares-sec-2010 Accessed 29 May 2013. 20 Instituto Nacional de Estadística (INE). Available from: http://www.ine.es; 2008 data. Accessed 29 May 2013. 21 El asma afecta a dos millones y medio de personas en España y su incidencia aumenta cada año. Patologías 19 April 2007. Available from: http://svadcf.es/noticia--_568 Accessed 29 May 2013. 22 El asma mal controlada produce el 70% del coste de la enfermedad en España. SEPAR, 08 May 2013. Available from: http://www.universidadpacientes.org/respiratorias/novedades/1557/ Accessed 29 May 2013. 23 El coste anual de la EPOC en España asciende a 3.000 millones de euros. SEPAR, 19 November 2012. Available from: http://www.universidadpacientes.org/respiratorias/novedades/1225/; Accessed 29 May 2013. 24 Jönsson, B., Staginnus, U. y W. Nils. La carga y el coste del cáncer en España. Economía de la Salud (ReES) 2007; 6 (3): 141-147. Available from: http://www.economiadelasalud.com/Ediciones/63/08_pdf/cargaycoste.pdf Accessed 29 May 2013. 25 Antoñanzas F, Oliva J, Velasco M, Zozaya N, Lorente R, Lopez-Bastida J. Costes directos e indirectos del cáncer en España. TRIBUNA DE ECONOMÍA. CUADERNOS ECONÓMICOS DE ICE N.º72. 281-309. Available from: http://www.revistasice.info/cachepdf/CICE_72_281309__B57A8C5C8944A31B899369D62E957DE4.pdf Accessed 29 May 2013. 26 Ministerio de Sanidad y Consumo Infección por VIH y SIDA en España. Plan multisectorial. Indicadores 2006. Available from: 235 http://www.msc.es/ciudadanos/enfLesiones/enfTransmisibles/sida/planesEstrat/ indicadores2006_3.pdf Accessed 29 May 2013. 27 World Health Organization. Available from: www.who.int Accessed 12 August 2013. 28 Instituto Nacional de Estadística. Available from: www.ine.es Accessed 12 August 2013. 29 Institut d’Estadística de Catalunya. Available from: www.idescat.cat Accessed 12 August 2013. 30 O'Sullivan AK, Thompson D, Drummond MF. Collection of health-economic data alongside clinical trials: is there a future for piggyback evaluations? Value Health. 2005;8:67–79. 31 Hlatky MA, Owens DK, Sanders GD. Cost-effectiveness as an outcome in randomized clinical trials. Clin Trials. 2006;3:543–551. 32 Arnold RG, Kotsanos JG, Motheral B, Ramsey S, Crown W, Puder K, Hornbrook M, Wright A, Murray M. Panel 3: Methodological Issues in Conducting Pharmacoeconomic Evaluations—Retrospective and Claims Database Studies. Value in Health, 1993, 2: 82–87. 33 Evans, C., & Crawford, B. Expert judgement in pharmacoeconomic studies. Guidance and future use. PharmacoEconomics, 2000, 17(6), 545-553. 34 Barr JT, Schumacher GE. Using decision analysis to conduct pharmacoeconomic studies. In: Spilker B, editor. Quality of life and pharmacoeconomics in clinical trials. 2nd ed. Philadelphia (PA): LippincottRaven, 1996. 35 Nuijten MJ, Pronk MH, Brorens MJ, Hekster YA, Lockefeer JH, de Smet PA. Reporting format for economic evaluation: Part II: Focus on modelling studies. Pharmacoeconomics 1998, 14:259-268. 36 Evans C. The use of consensus methods and expert panels in pharmacoeconomic studies: practical applications and methodological shortcomings. Pharmacoeconomics 1997; 12 (2 Pt 1): 121-9. 37 Halpern MT, Luce BR, Brown RE, Geneste B. Health economic modelling practices: a suggested framework. Value Health 1998; 1 (2):131-47. 38 Dalkey NC, Helmer O. (1963). An experimental application of the Delphi method to the use of experts. Management Science, 9 (3), 458-467. 39 Section II: Improving adherence rates: guidance for countries. WHO 2003. Available from: http://www.who.int/chp/knowledge/publications/Adherence_section2.pdf Accessed 20 June 2013. 40 Medicines adherence. Involving patients in decisions about prescribed medicines and supporting adherence. NICE January 2009. Available from: http://www.nice.org.uk/nicemedia/live/11766/43042/43042.pdf Accessed 20 June 2013. 41 Johnson FR. Returning to the Fold: Conventional Utility versus Health Utility for Risk-Benefit Analysis. Issues Panel: Health outcomes approaches to riskbenefit analysis: how ready are they? 11th International Meetings of the International Society for Pharmacoeconomics and Outcomes Research. May 2006. Philadelphia, PA. 236 42 Giambiasi N, Frydman C, Smaili M. 1996. Event Driven Simulation with Fuzzy Delay. Artificial Intelligence, Simulation, and Planning in High Autonomy Systems. La Jolla, USA. 43 Shannon, R.E. Systems simulation: The art and science, Prentice-Hall, Englewood Cliffs, 1975. 44 J. Banks, “Principles of Simulation”, in J. Banks (Editor), Handbook of Simulation: Principles, Methodology, Advances, Applications, and Practice, John Wiley & Sons, New York NY, 1998, pp3-30. 45 Nance, R. E., 1993, “History of Discrete Event Simulation Programming Languages,” Proceedings of the Second ACM SIGPLAN History of Programming Languages Conference, Cambridge, MA, April 20-23, Reprinted in ACM SIGPLAN Notices, 28(3), pp.149 175. 46 Zeigler, B, Moon Y, Kim D, Kim D. DEVS-C++: A high performance modeling and simulation environment. Technical Report, Department of Electrical and Computer Engineering, University of Arizona. In Proceedings of 29t. Hawaii International Conference on System Sciences, Jan. 1996. 47 Bodrogi J, Kaló Z. Principles of pharmacoeconomics and their impact on strategic imperatives of pharmaceutical research and development. Br J Pharmacol. 2010 Apr;159(7):1367-73. 48 Canadian Coordinating Office for Health Technology Assessment (CCOHTA). Guidelines for Economic Evaluation of Pharmaceuticals: Canada. 3rd edn. March 2006. Available from: http://www.ispor.org/PEguidelines/source/HTAGuidelinesfortheEconomicEvalua tionofHealthTechnologies-Canada.pdf Accessed November 2012. 49 National Institute for Clinical Excellence (April 2004). Guide to the methods of technology appraisal. Available from: www.nice.org.uk/niceMedia/pdf/TAP_Methods.pdf Accessed November 2012 50 Tambour M, Zethraeus N, Johannesson M. A note on confidence intervals in cost-effectiveness analysis. Int J Technol Assess Health Care, 1998. 14(3): p. 467-71. 51 Black W. The CE plane: A graphic representation of cost-effectiveness. Medical Decision Making. 1990; 10:212–214. 52 Fenwick E, Marshall DA, Levy AR, Nichol G. Using and interpreting costeffectiveness acceptability curves: an example using data from a trial of management strategies for atrial fibrillation. BMC Health Serv Res. 2006; 6: 52. Published online 2006 April 19. doi: 10.1186/1472-6963-6-52. 53 National Institute for Health and Clinical Excellence, 2008. Guide to the methods of technology appraisal [online]. Available from: http://www.nice.org.uk/media/B52/A7/ TAMethodsGuideUpdatedJune2008.pdf Accessed 20 June 2012. 54 Ortún V. 30.000 euros por AVAC. Econ Salud 2004, 17(49):1-2. 55 De Cock E, Miravitlles M, González-Juanatey JR, et al. Valor umbral del coste por año de vida ganado para recomendar la adopción de tecnologías sanitarias en España: evidencias procedentes de una revisión de la literatura. Pharmacoeconomics Spanish Research Articles 2007; 4: 97–107. 237 56 Briggs AH, Wonderling DE, Mooney CZ. Pulling cost-effectiveness analysis up by its bootstraps: a non-parametric approach to confidence interval estimation. Health Economics. 1997;6:327–340. 57 Lothgren M, Zethraeus N. Definition, interpretation and calculation of costeffectiveness acceptability curves. Health Economics. 2000;9:623–630. 58 Lynd L. Quantitative Methods for Therapeutic Risk-Benefit Analysis. Issues Panel: Health outcomes approaches to risk-benefit analysis: how ready are they? 11th International Meetings of the International Society for Pharmacoeconomics and Outcomes Research. May 2006. Philadelphia, PA. 59 Hauber AB, Johnson FR, Andrews EB. Risk-Benefit Analysis Methods for Pharmaceutical Decision-Making – Where Are We Now?. The Official News & Technical Journal Of The International Society For Pharmacoeconomics And Outcomes Research. Available from: https://www.ispor.org/news/articles/Dec06/RiskBenefitsAnalysis.asp Accessed 20 June 2012. 60 Edmunds WJ, Medley GF, Nokes DJ. Evaluating the cost-effectiveness of vaccination programmes: a dynamic perspective. Stat Med, 1999. 18(23): p. 3263-82. 61 Van de Velde N, Brisson M, Boily MC. Modeling human papillomavirus vaccine effectiveness: quantifying the impact of parameter uncertainty. Am J Epidemiol, 2007. 165(7): p. 762-75. 62 Yang M, Rajan S, Issa AM. Cost effectiveness of gene expression profiling for early stage breast cancer. Wiley Subscription Services, Inc., A Wiley Company Available from: http://onlinelibrary.wiley.com/doi/10.1002/cncr.27443/abstract Accessed 20 June 2012. 63 Ganz DA, Kuntz KM, Jacobson GA, Avorn J. Cost-effectiveness of 3hydroxy-3-methylglutaryl coenzyme A reductase inhibitor therapy in older patients with myocardial infarction. Ann Intern Med, 2000. 132(10): p. 780-7. 64 Schleinitz MD, Weiss JP, Owens DK. Clopidogrel versus aspirin for secondary prophylaxis of vascular events: a cost-effectiveness analysis. Am J Med., 2004. 116(12). 65 Goldie SJ, Paltiel AD, Weinstein MC, Losina E, Seage GR 3rd, Kimmel AD, Walensky RP, Sax PE, Freedberg KA. Projecting the cost-effectiveness of adherence interventions in persons with human immunodeficiency virus infection. Am J Med, 2003. 115(8): p. 632-41. 66 Schackman BR, Goldie SJ, Weinstein MC, Losina E, Zhang H, Freedberg KA. Cost-effectiveness of earlier initiation of antiretroviral therapy for uninsured HIV-infected adults. Am J Public Health, 2001. 91(9): p. 1456-63. 67 Walensky RP, Weinstein MC, Yazdanpanah Y, Losina E, Mercincavage LM, Touré S, Divi N, Anglaret X, Goldie SJ, Freedberg KA, the CEPAC-International Investigators. HIV drug resistance surveillance for prioritizing treatment in resource-limited settings. AIDS, 2007. 21(8): p. 973-82. 68 Morris BL, Scott CA, Wilkin TJ, Sax PE, Gulick RM, Freedberg KA, Schackman BR. Cost-effectiveness of Adding an Agent That Improves Immune Responses to Initial Antiretroviral Therapy (ART) in HIV-Infected Patients: Guidance for Drug Development. HIV Clin Trials. 2012 Jan-Feb; 13(1): 1–10. 238 69 Sanders GD, Bayoumi AM, Sundaram V, Bilir SP, Neukermans CP, Rydzak CE, Douglass LR, Lazzeroni LC, Holodniy M, Owens DK. Cost-effectiveness of screening for HIV in the era of highly active antiretroviral therapy. N Engl J Med, 2005. 352(6): p. 570-85. 70 Sax PE, Islam R, Walensky RP, Losina E, Weinstein MC, Goldie SJ, Sadownik SN, Freedberg KA. Should resistance testing be performed for treatment-naive HIV-infected patients? A cost-effectiveness analysis. Clin Infect Dis, 2005. 41(9): p. 1316-23. 71 Greeley SAW, John PM, Winn AN, Ornelas J, Lipton RB, Philipson LH, Bell GI, Huang ES. The Cost-Effectiveness of Personalized Genetic Medicine: The case of genetic testing in neonatal diabetes. Diabetes Care. 2011 March; 34(3): 622– 627. 72 Griffiths TL, Phillips CJ, Davies S, Burr ML, Campbell IA. Cost effectiveness of an outpatient multidisciplinary pulmonary rehabilitation programme. Thorax, 2001. 56(1): p. 779-84. 73 Hamel MB, Phillips RS, Davis RB, Desbiens N, Connors AF Jr, Teno JM, Wenger N, Lynn J, Wu AW, Fulkerson W, Tsevat J. Outcomes and costeffectiveness of initiating dialysis and continuing aggressive care in seriously ill hospitalized adults. SUPPORT Investigators. Study to Understand Prognoses and Preferences for Outcomes and Risks of Treatments. Ann Intern Med, 1997. 127(3): p. 195-202. 74 Zitzmann NU, Marinello CP, Sendi P. A cost-effectiveness analysis of implant overdentures. J Dent Res, 2006. 85(8): p. 717-21. 75 Edwards MJ, Brickley MR, Goodey RD, Shepherd JP. The cost, effectiveness and cost-effectiveness of removal and retention of asymptomatic, disease free third molars. British Dental Journal, 1999. 187(7): p. 380-384. 76 Dasbach EJ, Elbasha EH, Insinga RP. Mathematical models for predicting the epidemiologic and economic impact of vaccination against human papillomavirus infection and disease. Epidemiol Rev, 2006. 28: p. 88-100. 77 AIDS epidemic update. World Health Organization. Available from: http://www.unaids.org/en/media/unaids/contentassets/documents/epidemiology/ 2012/gr2012/20121120_UNAIDS_Global_Report_2012_en.pdf Accessed 13 March 2013. 78 United Nations Programme on HIV/AIDS and World Health Organization (UNAIDS/WHO). Country information. Available from: http://www.unaids.org/en/regionscountries/countries/spain/ Accessed 13 March 2013. 79 Palella FJ Jr, Delaney KM, Moorman AC, Loveless MO, Fuhrer J, Satten GA, Aschman DJ, Holmberg SD. Declining morbidity and mortality among patients with advanced human immunodeficiency virus infection. HIV Outpatient Study Investigators . N Engl J Med . 1998;338:853-60. 80 Nachega JB, Marconi VC, van Zyl GU, Gardner EM, Preiser W, Hong SY, Mills EJ, Gross R. HIV treatment adherence, drug resistance, virologic failure: evolving concepts. Infect Disord Drug Targets. 2011 Apr;11(2):167-74. 81 Zethraeus N, Ben Sedrine W, Caulin F, Corcaud S, Gathon HJ, Haim M, Johnell O, Jönsson B, Kanis JA, Tsouderos Y, Reginster JY. Models for assessing the 239 cost-effectiveness of the treatment and prevention of osteoporosis. Osteoporos Int, 2002. 13(11): p. 841-57. 82 Breiman L. Probability and stochastic processes: with a view towars applications (Boston: Houghton Mifflin, 1969). 83 Darbà, J, Restovic G. Class Lecture. Curso avanzado de farmacoeconomía. Universitat de Barcelona, Barcelona, Spain. 18-19 June 2007. 84 Drummond M, McGuire A. Economic evaluations in health care. Merging theory with practice. New York Oxford University Press, 2001. 85 Xin S. Markov Modelling in Healthcare Economic Evaluations. Chin J Evidbased Med, 2007, 7(10): 750-756. 86 Freedberg KA, Losina E, Weinstein MC, Paltiel AD, Cohen CJ, Seage GR, Craven DE, Zhang H, Kimmel AD, Goldie SJ. The cost effectiveness of combination antiretroviral therapy for HIV disease. N Engl J Med. 2001 Mar 15;344(11):824-31. 87 Fenwick E, Byford S. A guide to cost-effectiveness acceptability curves. Br J Psychiatry. 2005 Aug;187:106-8. 88 NIH Consensus Development Panel, Osteoporosis prevention, diagnosis, and therapy, JAMA 2001; 285:785-95. 89 Alldredge BK, Koda-Kimble MA, Young YL, Kradjan WA, Guglielmo BJ. Applied therapeutics: the clinical use of drugs, Wolters Kluwer Health/Lippincott Williams & Wilkens, Philadelphia, PA, 2009, pp. 101-3. 90 Sosa M, Gómez de Tejada MJ, Hernández Hernández D. Concepto, clasificación, factores de riesgo y clínica de la osteoporosis, Rev Esp Enf Metab Oseas 2001;10:7-11. 91 Melton LJ III, Atkinson EJ, Cooper C, O'Fallon WM, Riggs BL. Vertebral fractures predict subsequent fractures, Osteoporos Int 1999;10:214-21. 92 Bouza C, López T, Palma M, Amate JM. Hospitalised osteoporotic vertebral fractures in Spain: analysis of the national hospital discharge registry, Osteoporos Int 2007;18:649-57. 93 Sociedad Española de Investigaciones Óseas y Metabolismo Mineral (SEIOMM). Available from: http://www.seiomm.org/documentos/osteoporosis_es_en.pdf Accessed 13 March 2011. 94 Cooper C, Campion G, Melton LJ 3rd. Hip fractures in the elderly: a worldwide projection, Osteoporos Int 1992;2:285-9. 95 MacLean C, Newberry S, Maglione M, McMahon M, Ranganath V, Suttorp M, Mojica W, Timmer M, Alexander A, McNamara M, Desai SB, Zhou A, Chen S, Carter J, Tringale C, Valentine D, Johnsen B, Grossman J. Systematic review: comparative effectiveness of treatments to prevent fractures in men and women with low bone density or osteoporosis. Ann Intern Med 2008;148:197-213. 96 Lewiecki EM. Emerging drugs for postmenopausal osteoporosis. Expert Opin Emerg Drugs 2009;14:129-44. 97 Ettinger B, Black DM, Mitlak BH, Knickerbocker RK, Nickelsen T, Genant HK, Christiansen C, Delmas PD, Zanchetta JR, Stakkestad J, Glüer CC, Krueger K, Cohen FJ, Eckert S, Ensrud KE, Avioli LV, Lips P, Cummings SR. Reduction of vertebral fracture risk in postmenopausal women with osteoporosis treated with raloxifene: results from a 3-year randomized clinical trial. Multiple Outcomes of 240 Raloxifene Evaluation (MORE) Investigators. JAMA. 1999 Aug 18;282(7):63745. Erratum in: JAMA 1999 Dec 8;282(22):2124. 98 Arun B, Anthony M, Dunn B. The search for the ideal SERM. Expert Opin Pharmacother 2002;3:681-91. 99 Miller PD, Chines AA, Christiansen C, Hoeck HC, Kendler DL, Lewiecki EM, Woodson G, Levine AB, Constantine G, Delmas PD. Effects of bazedoxifene on BMD and bone turnover in postmenopausal women: 2-yr results of a randomized, double-blind, placebo-, and active-controlled study. J Bone Miner Res 2008;23:525-35. 100 Pinkerton JV, Archer DF, Utian WH, Menegoci JC, Levine AB, Chines AA, Constantine GD. Bazedoxifene effects on the reproductive tract in postmenopausal women at risk for osteoporosis. Menopause 2009;16:1102-8. 101 Del Pino Montes J. Epidemiología de las fracturas osteoporóticas: las fracturas vertebrales y no vertebrales. Rev Osteoporos Metab Miner 2010;2:S8-S12. 102 Ström O, Borgström F, Sen SS, Boonen S, Haentjens P, Johnell O, Kanis JA. Cost-effectiveness of alendronate in the treatment of postmenopausal women in 9 European countries--an economic evaluation based on the fracture intervention trial. Osteoporos Int 2007;18:1047-61. 103 Instituto de Salud Carlos III, Análisis coste-utilidad de los tratamientos farmacológicos para la prevención de fracturas en mujeres con osteoporosis en ESPAÑA. Informe Público de Evaluación de Tecnologías Sanitarias IPE 63/2010. Available from: http://www.isciii.es/htdocs/publicaciones/documentos/63 Osteoporosis Coste-utilidad farmacos. Accessed 13 May 2011. 104 Darba J, Restovic G, Kaskens L, Balbona MA, Carbonell A, Cavero P, Jordana M, Prieto C, Molina A, Padro I. Patient preferences for osteoporosis in Spain: a discrete choice experiment. Osteoporos Int 2011;22:1947-54. 105 Borgström F, Ström O, Kleman M, McCloskey E, Johansson H, Odén A, Kanis JA. Cost-effectiveness of bazedoxifene incorporating the FRAX(R) algorithm in a European perspective. Osteoporos Int 2011;22:955-65. 106 Sanfélix-Genovés J, Peiró S, Sanfélix-Gimeno G, Giner V, Gil V, Pascual M, Fluixá C, Fuertes A, Hurtado I, Ferreros I. Development and validation of a population-based prediction scale for osteoporotic fracture in the region of Valencia, Spain: the ESOSVAL-R study. BMC Public Health 2010;10:153. 107 Ismail AA, Pye SR, Cockerill WC, Lunt M, Silman AJ, Reeve J, Banzer D, Benevolenskaya LI, Bhalla A, Bruges Armas J, Cannata JB, Cooper C, Delmas PD, Dequeker J, Dilsen G, Falch JA, Felsch B, Felsenberg D, Finn JD, Gennari C, Hoszowski K, Jajic I, Janott J, Johnell O, Kanis JA, Kragl G, Lopez Vaz A, Lorenc R, Lyritis G, Marchand F, Masaryk P, Matthis C, Miazgowski T, Naves-Diaz M, Pols HA, Poor G, Rapado A, Raspe HH, Reid DM, Reisinger W, Scheidt-Nave C, Stepan J, Todd C, Weber K, Woolf AD, O'Neill TW. Incidence of limb fracture across Europe: results from the European Prospective Osteoporosis Study (EPOS) Osteoporos Int 2002;13:565-71. 108 European Prospective Osteoporosis Study (EPOS) Group, Felsenberg D, Silman AJ, Lunt M, Armbrecht G, Ismail AA, Finn JD, Cockerill WC, Banzer D, Benevolenskaya LI, Bhalla A, Bruges Armas J, Cannata JB, Cooper C, Dequeker J, Eastell R, Felsch B, Gowin W, Havelka S, Hoszowski K, Jajic I, Janott J, 247 184 Games PA, Keselman HJ, Clinch JJ. Tests for homogeneity of variance in factorial designs. Psychological Bulletin, 86, 978-984. 185 Sociedad Navarra de Medicina de Familia y Atención Primaria (SNAMFAP). Documento para el Manejo de la Osteoporosis en Atención Primaria (Actualización Diciembre 2006) Available from: http://www.guiasalud.es/GPC/GPC_363.pdf Accessed 16 February 2011. 186 Briggs AH, Wonderling DE, Mooney CZ. Pulling cost-effectiveness analysis up by its bootstraps: a non-parametric approach to confidence interval estimation. Health Economics. 1997;6:327–340. 187 Lothgren M, Zethraeus N. Definition, interpretation and calculation of costeffectiveness acceptability curves. Health Economics. 2000;9:623–630. 188 Rodríguez Barrios JM, Serrano D, Monleón T, Caro J. Los modelos de simulación de eventos discretos en la evaluación económica de tecnologías y productos sanitarios. Gaceta sanitaria: Organo oficial de la Sociedad Española de Salud Pública y Administración Sanitaria, ISSN 0213-9111, Vol. 22, Nº. 2, 2008, págs. 151-161. 189 Banks J, Georgia M. Discrete event simulation. Initially published in the Proceedings of the 1999 Winter Simulation Conference (ed. P.A. Farrington, H.B. Nembhard, D.T. Sturrock, G.W. Evans) pp. 7-13. Available from: http://www.telecom.otago.ac.nz/tele302/ref/Banks_DES.pdf Accessed 8 October 2013. 190 Overeinder B, Hertzberger B, Sloot P. Parallel Discrete Event Simulation. Abril 1991. Available from: http://dare.uva.nl/document/2069. Accessed 8 October 2013. 191 Field MJ, Gold GM, eds. (1998). Summarizing population health: Directions for the development and application of population metrics. Institute of Medicine, Washington, D.C.: National Academy Press. 192 Murray CJL, Salomon JA, Mathers C (2000). A critical examination of summary measures of population health. Bulletin of the World Health Organization 78 (8), 2000: 981-994. 193 Mathers CD, Vos T, Lopez AD, Salomon J, Ezzati M (ed.) 2001. National Burden of Disease Studies: A Practical Guide. Edition 2.0. Global Program on Evidence for Health Policy. Geneva: World Health Organization. 194 Dempsey M. Decline in tuberculosis: the death rate fails to tell the entire story. Am Rev Tuberc. 1947 Aug;56(2):157-64. 195 Virgili G, Koleva D, Garattini L, Banzi R, Gensini GF. Utilities and QALYs in health economic evaluations: glossary and introduction. Intern Emerg Med. 2010 Aug;5(4):349-52. 196 Murray CJL, Lopez, AD, eds. (1996a). The global burden of disease: a comprehensive assessment of mortality and disability from diseases, injuries and risk factors in 1990 and projected to 2020. Global Burden of disease and Injury Series, Vol. 1. Cambridge: Harvard University Press; 1996. 197 Sassi F. How to do (or not to do). Calculating QALYs, comparing QALY and DALY calculations. Oxford: The London School of Hygiene and Tropical Medicine. Oxford University Press. 2006:402-408. 248 198 Walley T. Pharmacoeconomics and Economic Evaluation of Drug Therapies. Available from: http://www.iuphar.org/pdf/hum_67.pdf Accessed 8 October 2013. 199 Medical Outcomes Trust (2007) MOS-HIV Health Survey (MOS-HIV). Available from: http://chipts.ucla.edu/assessment/pdf/as sessments/MOSHIV.35item.pdf Accessed 31 January 2007. 200 Marín F, Gonzalez-Macias J, Moya R, Onrubia C, Cancelo C, Alvarez S, Montero JJ, Gómez F, Carbonell C, Vila J, Díez-Pérez A, ECOSAP. Fragility non-spinal fractures in a cohort of 5,201 women aged 65 years and older during a 3-year follow-up. Med Clin (Barc) 2006;127:401-4. 201 Cooper C, Atkinson EJ, Jacobsen SJ, O’Fallon WM, Melton LJ III. Populationbased study of survival after osteoporotic fractures. Am J Epidemiol 1993;137:1001-5. 202 Cauley JA, Thompson DE, EnsrudKC, Scott JC, Black D. Risk of mortality following clinical fractures. Osteoporos Int 2000;11:556-61. 203 Center JR, Nguyen TV, Schneider D, Sambrook PN, Eisman JA. Mortality after all major types of osteoporotic fracture in men and women: an observational study. Lancet 1999;353:878-82. 204 Bliuc D, Nguyen ND, Milch VE, Nguyen TV, Eisman JA, Center JR. Center, Mortality risk associated with low-trauma osteoporotic fracture and subsequent fracture in men and women. JAMA 2009;301:513-21. 205 Agència d´Informació, Avaluació i Qualitat en Salut (AIAQS). Guía de Práctica Clínica sobre Osteoporosis y Prevención de Fracturas por Fragilidad. 2010. Guías de Práctica Clínica en el SNS: AATRM Nº 2007/02. Available from: http://www.gencat.cat/salut/depsan/units/aatrm/pdf/gpc_osteoporosi_aatrm2010 _vcompleta.pdf Accessed 13 May 2011. 206 Sociedad Navarra de Medicina de Familia y Atención Primaria (SNAMFAP), Documento para el Manejo de la Osteoporosis en Atención Primaria (Actualización Diciembre 2006). Available from: http://www.guiasalud.es/GPC/GPC_363.pdf. Accessed 29 April 2011. 207 Christodoulou C, Cooper C. What is osteoporosis? Postgrad Med J 2003;79:133-8. 208 Naves M, az-Lopez JB, Gomez C, Rodriguez-Rebollar A, Rodriguez-Garcia M, Cannata-Andia JB. The effect of vertebral fracture as a risk factor for osteoporotic fracture and mortality in a Spanish population. Osteoporos Int 2003;14:520-4. 209 Pinto JL, Sanchez F, Métodos para la evaluación económica de nuevas prestaciones. Editado por: Centre de Recerca en Economía i Salut – Cres y Ministerio de Sanidad y Consumo, España. Available from: http://www.msc.es. Accessed 29 April 2011. 210 Vademecum.es. Available from: http://www.vademecum.es/ Accessed 29 April 2011. 211 Borgstrom F, Jonsson B, Strom O, Kanis JA. An economic evaluation of strontium ranelate in the treatment of osteoporosis in a Swedish setting: based on the results of the SOTI and TROPOS trials. Osteoporos Int 2006;17:1781-93. 212 Jonsson B, Christiansen C, Johnell O, Hedbrandt J. Cost-effectiveness of fracture prevention in established osteoporosis. Osteoporos Int 1995;5:136-42. 249 213 Hospital Lluís Alcanyis, Memoria 2000. Available from: http://www.ctv.es/USERS/vgisbertj/home.html. Accessed 3 May 2011. 214 Hospital de la Esperanza, Tarifario 1995, Hospital de la Esperanza, Barcelona, 1995. 215 Instituto Nacional de la Salud, Resultados de la gestión analítica en los hospitales del INSALUD GECLIF 2000. Subdirección General de Coordinación Administrativa, Hospitales INSALUD, Madrid, 2001. 216 Finnern HW, Sykes DP. The hospital cost of vertebral fractures in the EU: estimates using national datasets. Osteoporos Int 2003;14:429-36. 217 Cernuda C. Coste de consultas externas: visitas y pruebas. Todo Hospital, nº 153 Enero/Febrero, 1999, H. Universitario Josep Trueta, Girona, 1999. 218 Diari Oficial de la Generalitat de Catalunya (DOGC). Departament de Sanitat i Seguretat Social. Orden de 29 de Septiembre de 1997; núm. 2504. 219 de Villiers TJ, Chines AA, Palacios S, Lips P, Sawicki AZ, Levine AB, Codreanu C, Kelepouris N, Brown JP. Safety and tolerability of bazedoxifene in postmenopausal women with osteoporosis: results of a 5-year, randomized, placebo-controlled phase 3 trial. Osteoporos Int 2011;22:567-76. 220 Diari Oficial de la Generalitat de Catalunya (DOGC). Institut Català de la Salut. Departament de Salut. RESOLUCIÓN SLT/383/2009, de 21 de enero, sobre la revisión de precios públicos correspondientes a los servicios sanitarios que presta el Instituto Catalán de la Salud. Febrero 2009; núm. 5325, Available from: https://www.gencat.cat/eadop/imagenes/5325/09042029.pdf. Accessed 3 March 2011. 221 Boletín Oficial De La Rioja (BOR). Consejería de Salud. Resolución del Consejero de Salud por la que se dispone la publicación de las tarifas por servicios sanitarios prestados a particulares en los centros del Servicio Riojano de Salud. Febrero 2009; núm. 27. Available from: http://www2.larioja.org/pls/dad_user/G04.texto_integro?p_cdi_accn=443230928. Accessed 3 March 2011. 222 Adachi JD, Adami S, Gehlbach S, Anderson FA Jr, Boonen S, Chapurlat RD, Compston JE, Cooper C, Delmas P, Díez-Pérez A, Greenspan SL, Hooven FH, LaCroix AZ, Lindsay R, Netelenbos JC, Wu O, Pfeilschifter J, Roux C, Saag KG, Sambrook PN, Silverman S, Siris ES, Nika G, Watts NB; GLOW Investigators. Impact of prevalent fractures on quality of life: baseline results from the global longitudinal study of osteoporosis in women. Mayo Clin Proc 2010;85:806-13. 223Sobocki P, Lekander I, Borgstrom F, Strom O, Runeson B. The economic burden of depression in Sweden from 1997 to 2005. Eur Psychiatry 2007;22:146-52. 224 Zethraeus N, Borgstrom F, Jonsson B, Kanis J. Reassessment of the costeffectiveness of hormone replacement therapy in Sweden: results based on the Women's Health Initiative randomized controlled trial. Int J Technol Assess Health Care 2005;21:433-41. 225 Sanders GD, Bayoumi AM, Sundaram V, et al. Cost-effectiveness of screening for HIV in the era of highly active antiretroviral therapy. N Engl J Med 2005; 352:570-85. 226 Anders B, Ommen O, Pfaff H, Lüngen M, Lefering R, Thüm S, Janssen C. Direct, indirect, and intangible costs after severe trauma up to occupational 250 reintegration - an empirical analysis of 113 seriously injured patients. Psychosoc Med. 2013 Jun 17;10:Doc02. doi: 10.3205/psm000092. Print 2013. 227 National Statistics Institute: Instituto Nacional de Estadística. Indice de precios al consumo. Available from http://www.ine.es/calcula/index.do?L=0 Accessed 14 June 2011. 228 Ware JE, Sherbourne CD. The MOS 36-item short-form health survey (SF36). I. Conceptual framework and item selection. Med Care 1992; 30: 473-483. 229 Ware JE Jr, Kosinski M, Keller SD. A 12-Item Short-Form Health Survey: construction of scales and preliminary tests of reliability and validity. Med Care. 1996; 34:220-33. 230 Ware JE, Kosinski M, Keller SD. How to Score the SF-12® Physical and Mental Health Summary Scales. 3rd ed. Lincoln (RI): QualityMetric; 1998. 251 ADDENDA 252 253 Addendum I: Publications related with this thesis as of June 2014 1. Darbà J, Kaskens L, Garreta A, Paredes R, Pérez-Álvarez N. The economic and clinical consequences of pre-treatment human and viral genotyping screening for antiretroviral treatment in HIV infected patients. Pharmacoeconomics. Under revivew. 2. Darbà J, Kaskens L, Pérez-Álvarez N, Palacios S, Neyro JL, Rejas J. Disability-adjusted-life-year loss in postmenopausal women with osteoporosis: a burden of illness study. Menopause. Under review. 3. Darbà J, Pérez-Álvarez N, Kaskens L, Holgado-Pérez S, Racketa J, Rejas J. Cost-effectiveness of bazedoxifene versus raloxifene in the treatment of postmenopausal women in Spain. Clinicoecon Outcomes Res. 2013 Jul 5; 5:327-36. 254 255 Addendum II: Conference contributions related with this thesis as of June 2014 Invited presentation 1. ‘Markov models used in a 2-stage outcome cohort simulation for an economic evaluation’. 5th International Conference of the ERCIM WG on Computing & Statistics (ERCIM 2012). 1-3 December 2012. Oviedo, Spain. 2. Perez-Alvarez N, Gomez G, Paredes R, Clotet B. Costeffectiveness of HIV tropism testing to inform antiretroviral treatment with maraviroc. 6th meeting of the Eastern Mediterranean Region International Biometric Society (EMRIBS). 8-12 May 2011, Crete, Greece. Contributed presentations 1. Pérez-Álvarez N, Muñoz-Moreno JA, Gomez G. Cost effectiveness evaluation for promoting HIV treatment adherence: cohort simulation using a pilot study data. 7th meeting of the Eastern Mediterranean Region International Biometric Society (EMR-IBS). 22-25 April, 2013. Tel Aviv, Israel. 2. Pérez-Álvarez N, Kaskens L, Darbà J. Cost-effectiveness study of treatments for fracture prevention in postmenopausal women. 2ª Reunión General Biostatnet. January 25-26 2013. Santiago de Compostela, Spain. Posters 1. Darba J, Kaskens L, Pérez-Álvarez N, Palacios S, Neyro JL, Rejas J. Disability-adjusted life years loss in postmenopausal women receiving major pharmacological interventions for osteoporosis. International Osteoporosis Foundation, European 256 Congress on Osteoporosis and Osteoarthritis, Orthopaedics Medical Congress. 17 – 20 April 2013. Rome, Italy. 2. Muñoz-Moreno JA, Gillen-Marconi M, Pérez-Álvarez N, Fumaz CR, González-García M, Ferrer MJ, Clotet B.Promotion of HIV Treatment Adherence and its Economical Cost: A Preliminary Cost-Effectiveness Analysis from a Controlled Randomized Prospective Trial. 6th IAS Conference on HIV pathogenesis, treatment and prevention. 17-20 July 2011, Rome, Italy. Posters with material not included in the dissertation but which constitute an example of application of the methods tackled in the thesis are: 1. Darbà J, Kaskens L, Pérez-Álvarez N. Neuropathic pain: a budget impact analysis to estimate costs due to the introduction of Qutenza® on the spanish market. ISPOR 14th annual European Congress (International Society for Pharmacoeconomics and Outcomes Research). 5-8 November, 2011, Madrid, Spain. 2. Darbà J, Kaskens L, Pérez-Álvarez N. Cost analysis of haemostatic treatment with a fibrin-based sponge versus fibrin sealant in lung surgery and liver resection in a spanish setting. ISPOR 14th annual European Congress (International Society for Pharmacoeconomics and Outcomes Research). 5-8 November, 2011, Madrid, Spain. 3. Darbà J, Pérez-Álvarez N, Kaskens L, Martín P. Dasatinib or imatinib in newly diagnosed chronic myeloid leukaemia patients in the chronic phase: Five-years follow-up simulated cohort. ISPOR 14th annual European Congress (International Society for Pharmacoeconomics and Outcomes Research). 5-8 November, 2011, Madrid, Spain. 263 MULTINEKA Study Group. Improvement of mitochondrial toxicity in patients receiving a nucleoside reverse-transcriptase inhibitorsparing strategy: results from the Multicenter Study with Nevirapine and Kaletra (MULTINEKA). Clin Infect Dis. 2009 Sep 15; 49(6):892-900. 34. Mothe B, Perez I, Domingo P, Podzamczer D, Ribera E, Curran A, Viladés C, Vidal F, Dalmau D, Pedrol E, Negredo E, Moltó J, Paredes R, Perez-Alvarez N, Gatell JM, Clotet B. HIV-1 infection in subjects older than 70: a multicenter cross-sectional assessment in Catalonia, Spain. Curr HIV Res. 2009 ;7(6):597-600. 35. Fumaz CR, Muñoz-Moreno JA, Ferrer MJ, Negredo E, PérezAlvarez N, Tarrats A, Clotet B. Low levels of adherence to antiretroviral therapy in HIV-1-infected women with menstrual disorders. AIDS Patient Care STDS. 2009 Jun; 23(6):463-8. 36. Negredo E, Puig J, Aldea D, Medina M, Estany C, Pérez-Alvarez N, Rodríguez-Fumaz C, Muñoz-Moreno JA, Higueras C, GonzalezMestre V, Clotet B. Four-year safety with polyacrylamide hydrogel to correct antiretroviral-related facial lipoatrophy. AIDS Res Hum Retroviruses. 2009 Apr; 25(4):451-5. 37. Tural C, Tor J, Sanvisens A, Pérez-Alvarez N, Martínez E, Ojanguren I, García-Samaniego J, Rockstroh J, Barluenga E, Muga R, Planas R, Sirera G, Rey-Joly C, Clotet B. Accuracy of simple biochemical tests in identifying liver fibrosis in patients coinfected with human immunodeficiency virus and hepatitis C virus. Clin Gastroenterol Hepatol. 2009 Mar; 7(3):339-45. 38. Moltó J, Santos JR, Pérez-Alvarez N, Cedeño S, Miranda C, Khoo S, Else L, Llibre JM, Valle M, Clotet B. Darunavir inhibitory quotient predicts the 48-week virological response to darunavirbased salvage therapy in human immunodeficiency virus-infected 264 protease inhibitor-experienced patients. Antimicrob Agents Chemother. 2008 Nov; 52(11):3928-32. 39. Bonjoch A, Buzon MJ, Llibre JM, Negredo E, Puig J, Pérez-Alvarez N, Videla S, Martinez-Picado J, Clotet B. Transient treatment exclusively containing nucleoside analogue reverse transcriptase inhibitors in highly antiretroviral-experienced patients preserves viral benefit when a fully active therapy was initiated. HIV Clin Trials. 2008 Nov-Dec; 9(6):387-98. 40. Muñoz-Moreno JA, Fumaz CR, Ferrer MJ, Prats A, Negredo E, Garolera M, Pérez-Alvarez N, Moltó J, Gómez G, Clotet B. Nadir CD4 cell count predicts neurocognitive impairment in HIV-infected patients. AIDS Res Hum Retroviruses. 2008 Oct; 24(10):1301-7. 41. Llibre JM, Bonjoch A, Iribarren J, Galindo MJ, Negredo E, Domingo P, Pérez-Alvarez N, Martinez-Picado J, Schapiro J, Clotet B; HIV Conference Call Study Group. Targeting only reverse transcriptase with zidovudine/lamivudine/abacavir plus tenofovir in HIV-1infected patients with multidrug-resistant virus: a multicentre pilot study. HIV Med. 2008, Aug; 9(7):508-13. 42. Negredo E, Puigdomènech I, Marfil S, Puig J, Pérez-Álvarez N, Ruiz L,Rey-Joly C, Clotet B, Blanco J. Association between HIV replication and cholesterol in peripheral blood mononuclear cells in HIV-infected patients interrupting HAART. J Antimicrob Chemother. 2008 Feb; 61(2): 400-4. 43. Llibre JM, Perez-Alvarez N. Hill A, Moyle G. Relative antiviral efficacy of ritonavir-boosted darunavir and ritonavir-boosted tipranavir vs. control protease inhibitor in the POWER and RESIST trials. HIV Med 2007; 8: 259-264. Methodological accuracy in crosstrial comparisons of antiretroviral regimens in multitreated patients. HIV Med. 2007, Nov; 8(8):568-70.