Search

Objectives Developing and validating a discrete event simulation model that is able to model patients with heart failure managed with usual care or an early warning system (with or without a diagnostic algorithm) and to account for the impact of individual patient characteristics in their health outcomes. Methods The model was developed using patient-level data from the Trans-European Network – Home-Care Management System study. It was coded using RStudio Version 1.3.1093 (version 3.6.2.) and validated along the lines of the Assessment of the Validation Status of Health-Economic decision models tool. The model includes 20 patient and disease characteristics and generates 8 different outcomes. Model outcomes were generated for the base-case analysis and used in the model validation. Results Patients managed with the early warning system, compared with usual care, experienced an average increase of 2.99 outpatient visits and a decrease of 0.02 hospitalizations per year, with a gain of 0.81 life years (0.45 quality-adjusted life years) and increased average total costs of €11 249. Adding a diagnostic algorithm to the early warning system resulted in a 0.92 life year gain (0.57 quality-adjusted life years) and increased average costs of €9680. These patients experienced a decrease of 0.02 outpatient visits and 0.65 hospitalizations per year, while they avoided being hospitalized 0.93 times. The model showed robustness and validity of generated outcomes when comparing them with other models addressing the same problem and with external data. Conclusions This study developed and validated a unique patient-level simulation model that can be used for simulating a wide range of outcomes for different patient subgroups and treatment scenarios. It provides useful information for guiding research and for developing new treatment options by showing the hypothetical impact of these interventions on a large number of important heart failure outcomes.

Objective The fields of medicine and public health are undergoing a data revolution. An increasing availability of data has brought about a growing interest in machine-learning algorithms. Our objective is to present the reader with an introduction to a knowledge representation and machine-learning tool for risk estimation in medical science known as Bayesian networks (BNs). Study Design In this article we review how BNs are compact and intuitive graphical representations of joint probability distributions (JPDs) that can be used to conduct causal reasoning and risk estimation analysis and offer several advantages over regression-based methods. We discuss how BNs represent a different approach to risk estimation in that they are graphical representations of JPDs that take the form of a network representing model random variables and the influences between them, respectively. Methods We explore some of the challenges associated with traditional risk prediction methods and then describe BNs, their construction, application, and advantages in risk prediction based on examples in cancer and heart disease. Results Risk modeling with BNs has advantages over regression-based approaches, and in this article we focus on three that are relevant to health outcomes research: (1) the generation of network structures in which relationships between variables can be easily communicated; (2) their ability to apply Bayes’s theorem to conduct individual-level risk estimation; and (3) their easy transformation into decision models. Conclusions Bayesian networks represent a powerful and flexible tool for the analysis of health economics and outcomes research data in the era of precision medicine.

Background The conditional reimbursement policy for expensive medicines in The Netherlands requires data collection on actual use and cost-effectiveness after the initial decision to reimburse a drug. This introduces new sources of uncertainty (less important in a randomized controlled trial than in daily practice), which may affect priorities for further research. Objectives This article focuses on determining the impact of including these uncertainties at the time a decision is made, and whether more complex models are always needed to address prioritization of additional research. Methods We constructed a typical decision model for chronic progressive diseases with four health states and parameters related to transition and exacerbation probabilities, costs, and utilities. Different scenarios are built on the basis of three additional uncertainties: persistence, compliance, and broadening of indication. Persistence refers to treatment duration. Compliance describes the fraction of treatment benefit obtained because of not taking the medication as prescribed. Broadening of indication reflects a shift in the severity distribution at treatment start. These uncertainties were parameterized in the model and included in the value-of-information analysis. Results The most important parameters were transition probabilities. Broadening of indication had little impact on the overall uncertainty. Compliance and persistence were important when establishing priorities for further research. Major differences with respect to the reference scenario were due to the parameterization of compliance in the decision model. Conclusions The usual practice of modeling only randomized controlled trial data at the time the decision on conditional reimbursement is made can lead to wrong decisions. Additional uncertainties arising from outcomes studies should be anticipated at an early stage and included in the model because this can have a strong impact on the prioritization of further research.

Background Acute myeloid leukemia (AML) can be diagnosed at any age and treatment, which can be given with supportive and/or curative intent, is considered expensive compared with that for other cancers. Despite this, no long-term predictive models have been developed for AML, mainly because of the complexities associated with this disease. Objective The objective of the current study was to develop a model (based on a UK cohort) to predict cost and life expectancy at a population level. Methods The model developed in this study combined a decision tree with several Markov models to reflect the complexity of the prognostic factors and treatments of AML. The model was simulated with a cycle length of 1 month for a time period of 5 years and further simulated until age 100 years or death. Results were compared for two age groups and five different initial treatment intents and responses. Transition probabilities, life expectancies, and costs were derived from a UK population-based specialist registry—the Haematological Malignancy Research Network (www.hmrn.org). Results Overall, expected 5-year medical costs and life expectancy ranged from £8,170 to £81,636 and 3.03 to 34.74 months, respectively. The economic and health outcomes varied with initial treatment intent, age at diagnosis, trial participation, and study time horizon. The model was validated by using face, internal, and external validation methods. The results show that the model captured more than 90% of the empirical costs, and it demonstrated good fit with the empirical overall survival. Conclusions Costs and life expectancy of AML varied with patient characteristics and initial treatment intent. The robust AML model developed in this study could be used to evaluate new diagnostic tools/treatments, as well as enable policy makers to make informed decisions.

Objectives This article proposes an integrated approach to the development, validation, and evaluation of new risk prediction models illustrated with the Fungal Infection Risk Evaluation study, which developed risk models to identify non-neutropenic, critically ill adult patients at high risk of invasive fungal disease (IFD). Methods Our decision-analytical model compared alternative strategies for preventing IFD at up to three clinical decision time points (critical care admission, after 24 hours, and end of day 3), followed with antifungal prophylaxis for those judged "high" risk versus "no formal risk assessment." We developed prognostic models to predict the risk of IFD before critical care unit discharge, with data from 35,455 admissions to 70 UK adult, critical care units, and validated the models externally. The decision model was populated with positive predictive values and negative predictive values from the best-fitting risk models. We projected lifetime cost-effectiveness and expected value of partial perfect information for groups of parameters. Results The risk prediction models performed well in internal and external validation. Risk assessment and prophylaxis at the end of day 3 was the most cost-effective strategy at the 2% and 1% risk threshold. Risk assessment at each time point was the most cost-effective strategy at a 0.5% risk threshold. Expected values of partial perfect information were high for positive predictive values or negative predictive values (£11 million–£13 million) and quality-adjusted life-years (£11 million). Conclusions It is cost-effective to formally assess the risk of IFD for non-neutropenic, critically ill adult patients. This integrated approach to developing and evaluating risk models is useful for informing clinical practice and future research investment.

Item does not contain fulltext OBJECTIVES: In early stages of development of new medical technologies, there are conceptually separate but related societal decisions to be made concerning adoption, further development (i.e., technical improvement), and research (i.e., clinical trials) of new technologies. This article presents a framework to simultaneously support these three decisions from a societal perspective. The framework is applied to the 70-gene signature, a gene-expression profile for breast cancer, deciding which patients should receive adjuvant systemic therapy after surgery. The "original" signature performed on fresh frozen tissue (70G-FFT) could be further developed to a paraffin-based signature (70G-PAR) to reduce test failures. METHODS: A Markov decision model comparing the "current" guideline Adjuvant Online (AO), 70G-FFT, and 70G-PAR was used to simulate 20-year costs and outcomes in a hypothetical cohort in The Netherlands. The 70G-PAR strategy was based on projected data from a comparable technology. Incremental net monetary benefits were calculated to support the adoption decision. Expected net benefit of development for the population and expected net benefit of sampling were calculated to support the development and research decision. RESULTS: The 70G-PAR had the highest net monetary benefit, followed by the 70G-FFT. The population expected net benefit of development amounted to euro91 million over 20 years (assuming euro250 development costs per patient receiving the test). The expected net benefit of sampling amounted to euro61 million for the optimal trial (n = 4000). CONCLUSIONS: We presented a framework to simultaneously support adoption, development, and research decisions in early stages of medical technology development. In this case, the results indicate that there is value in both further development of 70G-FFT into 70G-PAR and further research.

Objectives To investigate the importance of accounting for potential performance dependency when evaluating the cost-effectiveness of two diagnostic tests used in combination. Methods Two meta-analysis models were fitted to estimate the diagnostic accuracy of Wells score and Ddimer in combination. The first model assumes that the two tests perform independently of one another; thus, two separate meta-analyses were fitted to the Ddimer and Wells score data and then combined. The second model allows for any performance dependency of the two tests by incorporating published data on the accuracy of Ddimer stratified by Wells score, as well as studies of Ddimer alone and Wells score alone. The results from the two meta-analysis models were input into a decision model to assess the impact that assumptions regarding performance dependency have on the overall cost-effectiveness of the tests. Results The results highlight the importance of accounting for potential performance dependency when evaluating the cost-effectiveness of diagnostic tests used in combination. In our example, assuming the diagnostic performance of the two tests to be independent resulted in the strategy "Wells score moderate/high risk treated for DVT and Wells score low risk tested further with Ddimer" being identified as the most cost-effective at the £20,000 willingness-to-pay threshold (probability cost-effective 0.8). However, when performance dependency is modeled, the most cost-effective strategies were "Ddimer alone" and "Wells score low/moderate risk discharged and Wells score high risk further tested with Ddimer" (probability cost-effective 0.4). Conclusions When evaluating the effectiveness and cost-effectiveness of diagnostic tests used in combination, failure to account for diagnostic performance dependency may lead to erroneous results and nonoptimal decision making.

Background The expected value of partial perfect information (EVPPI) is a theoretically justifiable and informative measure of uncertainty in decision-analytic cost-effectiveness models, but its calculation is computationally intensive because it generally requires two-level Monte Carlo simulation. We introduce an efficient, one-level simulation method for the calculation of single-parameter EVPPI. Objective We show that under mild regularity assumptions, the expectation-maximization-expectation sequence in EVPPI calculation can be transformed into an expectation-maximization-maximization sequence. By doing so, calculations can be performed in a single-step expectation by using data generated for probabilistic sensitivity analysis. We prove that the proposed estimator of EVPPI converges in probability to the true EVPPI. Methods and Results The performance of the new method was empirically demonstrated by using three exemplary decision models. Our proposed method seems to achieve remarkably higher accuracy than the two-level method with a fraction of its computation costs, though the achievement in accuracy was not uniform and varied across the parameters of the models. Software is provided to calculate single-parameter EVPPI based on the probabilistic sensitivity analysis data. Conclusions The new method, though applicable only to single-parameter EVPPI, is fast, accurate, and easy to implement. Further research is needed to evaluate the performance of this method in more complex scenarios and to extend such a concept to similar measures of decision uncertainty.

Objective Decision models are sometimes used alongside systematic reviews to synthesize evidence. Clarity, however, is lacking about when and how to conduct modeling studies in tandem with systematic reviews, as well as about how to evaluate and present model results. The objective of this study was to collect and analyze information from various sources to inform the development of a framework for deciding when and how a decision model should be added to a systematic review. Methods We collected data through 1) review and analysis of evidence reports that used decision models; 2) review and synthesis of current best practices for the development of decision models; 3) interviews of Evidence-Based Practice Center directors and selected staff, United States Preventive Services Task Force members, and decision modelers who developed models used by the United States Preventive Services Task Force; and 4) a focus group of expert modelers. Results Models are well suited to address gaps in the literature, better suited for certain types of research questions, and essential for determining the value of information relating to future research. Opinions differ regarding whether model outputs constitute evidence, but interviewees expressed concern over the lack of standards and directions in grading and reporting such "evidence." Interviews of stakeholders and modelers revealed the importance of communication and presentation of model results as well as the importance of model literacy and involvement of stakeholders. Conclusions The study demonstrates the need for a framework for deciding when and how to use models alongside systematic reviews and provides information to develop such a framework.

A model's purpose is to inform medical decisions and health care resource allocation. Modelers employ quantitative methods to structure the clinical, epidemiological, and economic evidence base and gain qualitative insight to assist decision makers in making better decisions. From a policy perspective, the value of a model-based analysis lies not simply in its ability to generate a precise point estimate for a specific outcome but also in the systematic examination and responsible reporting of uncertainty surrounding this outcome and the ultimate decision being addressed. Different concepts relating to uncertainty in decision modeling are explored. Stochastic (first-order) uncertainty is distinguished from both parameter (second-order) uncertainty and from heterogeneity, with structural uncertainty relating to the model itself forming another level of uncertainty to consider. The article argues that the estimation of point estimates and uncertainty in parameters is part of a single process and explores the link between parameter uncertainty through to decision uncertainty and the relationship to value of information analysis. The article also makes extensive recommendations around the reporting of uncertainty, in terms of both deterministic sensitivity analysis techniques and probabilistic methods. Expected value of perfect information is argued to be the most appropriate presentational technique, alongside cost-effectiveness acceptability curves, for representing decision uncertainty from probabilistic analysis.

Objectives: The study aims to assess how evidence on diagnostic test accuracy is synthesized and used to inform economic decision modeling for HTA. Methods: All reports evaluating diagnostic test via an economic decision model published by the NHS Research and Development Health Technology Assessment (HTA) program since 1997 were identified. The methods for evidence synthesis of diagnostic test accuracy data and its use in economic decision modeling in this sample were reviewed. Results: Forty-four HTA reports out of 474 concerned diagnostic accuracy, of which 11 did not do any economic evaluation. Of the remaining 33 HTAs, 14 conducted meta-analyses of diagnostic accuracy in the clinical review but only 8 used such pooled estimates to inform the decision model. A number of meta-analysis methods ranging in complexity were applied to estimate diagnostic accuracy. Nevertheless, when it came to informing the economic decision model, the majority of reviews used independent meta-analytic estimates of sensitivity and specificity. Conclusions: Often, very simplistic methods to estimate diagnostic test accuracy were used for purposes of informing an economic decision model. The assumptions made by the simplistic methods are usually invalid which may lead to suboptimal decisions being made. It is desirable that decision modelers become aware of the rapid evolution of metaanalysis methods in this area; however, further research is still required to identify how the pooled results obtained from the different meta-analysis models should best be used to inform economic decision models.

Objectives Increased use of discrete choice experiments (DCEs) in healthcare requires establishing whether stated preferences are predictive of observed healthcare utilization. This study aimed to determine whether the number of alternatives in a DCE choice task should reflect the actual decision context, and how complex the choice model needs to be to predict real-world choices correctly at an aggregate and individual level in healthcare. Methods Two randomized controlled trials (RCTs) involving choices for influenza vaccination and colorectal cancer screening were used. Each RCT had three study conditions: DCE choice tasks with (i) two alternatives, (ii) three alternatives, or (iii) both. Two samples of 1,200 respondents each were randomly assigned to one of the conditions. Each respondent answered 16 DCE choice tasks (for the derivation of the decision model) plus a choice task mimicking the real-world choice (to keep the decision context the same). The data was analysed in a systematic way using random-utility-maximization (RUM) and random-regret-minimization (RRM) choice processes with scale and/or preference heterogeneity. Results Irrespective of the number of alternatives per choice task, the choice to opt for influenza vaccination and colorectal cancer screening was correctly predicted by DCE at an aggregate level, if scale and preference heterogeneity were taken into account. At an individual level, three alternatives per choice task and using heteroscedastic model plus preference heterogeneity seemed to be most promising, correctly predicting the real-world choice in 81.7% and 87.9% of the cases for influenza vaccination and colorectal cancer screening respectively. No evidence was found that RRM outperformed RUM. Conclusions Our study shows that DCEs hold the potential of being externally valid at an aggregate level if at least scale and preference heterogeneity are taken into account. Further research is needed to determine if this result remains in other contexts, and to optimise choice prediction at an individual level.

Background The methods used to estimate health-state utility values (HSUV) for multiple health conditions can produce very different values. Economic results generated using baselines of perfect health are not comparable with those generated using baselines adjusted to reflect the HSUVs associated with the health condition. Despite this, there is no guidance on the preferred techniques and little research describing the effect on cost per quality adjusted life-year (QALY) results when using the different methods. Methods Using a cardiovascular disease (CVD) model and cost per QALY thresholds, we assess the consequence of using different baseline health-state utility profiles (perfect health, no history of CVD, general population) in conjunction with models (minimum, additive, multiplicative) frequently used to approximate scores for health states with multiple health conditions. HSUVs are calculated using the EQ-5D UK preference-based algorithm. Results Assuming a baseline of perfect health ignores the natural decline in quality of life associated with age, overestimating the benefits of treatment. The results generated using baselines from the general population are comparable to those obtained using baselines from individuals with no history of CVD. The minimum model biases results in favor of younger-aged cohorts. The additive and multiplicative models give similar results. Conclusion Although further research in additional health conditions is required to support our findings, our results highlight the need for analysts to conform to an agreed reference case. We demonstrate that in CVD, if data are not available from individuals without the health condition, HSUVs from the general population provide a reasonable approximation.

In most decision-analytic models in health care, it is assumed that there is treatment without delay and availability of all required resources. Therefore, waiting times caused by limited resources and their impact on treatment effects and costs often remain unconsidered. Queuing theory enables mathematical analysis and the derivation of several performance measures of queuing systems. Nevertheless, an analytical approach with closed formulas is not always possible. Therefore, simulation techniques are used to evaluate systems that include queuing or waiting, for example, discrete event simulation. To include queuing in decision-analytic models requires a basic knowledge of queuing theory and of the underlying interrelationships. This tutorial introduces queuing theory. Analysts and decision-makers get an understanding of queue characteristics, modeling features, and its strength. Conceptual issues are covered, but the emphasis is on practical issues like modeling the arrival of patients. The treatment of coronary artery disease with percutaneous coronary intervention including stent placement serves as an illustrative queuing example. Discrete event simulation is applied to explicitly model resource capacities, to incorporate waiting lines and queues in the decision-analytic modeling example.

Background The characterization of uncertainty is critical in cost-effectiveness analysis, particularly when considering whether additional evidence is needed. In addition to parameter and methodological uncertainty, there are other sources of uncertainty which include simplifications and scientific judgments that have to be made when constructing and interpreting a model of any sort. These have been classified in a number of different ways but can be referred to collectively as structural uncertainties. Materials and Methods Separate reviews were undertaken to identify what forms these other sources of uncertainty take and what other forms of potential methods to explicitly characterize these types of uncertainties in decision analytic models. These methods were demonstrated through application to four decision models each representing one of the four types of uncertainty. Results These sources of uncertainty fall into four general themes: 1) inclusion of relevant comparators; 2) inclusion of relevant events; 3) alternative statistical estimation methods; and 4) clinical uncertainty. Two methods to explicitly characterize such uncertainties were identified: model selection and model averaging. In addition, an alternative approach, adding uncertain parameters to represent the source of uncertainty was also considered. The applications demonstrate that cost-effectiveness may be sensitive to these uncertainties and the methods used to characterize them. The value of research was particularly sensitive to these uncertainties and the methods used to characterize it. It is therefore important, for decision-making purposes, to incorporate such uncertainties into the modeling process. Conclusion Only parameterizing the uncertainty directly in the model can inform the decision to conduct further research to resolve this source of uncertainty.

Contains fulltext : 53088.pdf (Publisher’s version ) (Closed access) OBJECTIVE: To establish cost-effectiveness of antiepileptic drug (AED) treatment strategies of newly diagnosed patients with epilepsy. METHODS: A decision analysis was carried out comparing effectiveness and treatment cost of six treatment strategies comprising carbamazepine (CBZ), lamotrigine (LTG), and valproate (VPA) as first-line and second-line drugs. Three outcome groups were defined: complete success, partial success, and failure. Data on seizure control and failure due to adverse effects were derived from the literature. Data on resource use and costs were collected for each outcome group by means of a patient survey. RESULTS: Cost data were obtained from 71 patients. Cost increased from complete success to failure outcome groups. The probability of obtaining complete success varied from 64% (VPA-CBZ strategy) to 74% (LTG-VPA strategy). The strategy LTG-VPA was more effective than the least expensive strategy CBZ-VPA, but at higher costs per additional effectively treated patient. Probabilistic sensitivity analysis confirmed these findings to be robust. Subsequent analysis showed that changing inclusion criteria used in the selection of the studies from the literature had a major effect on cost-effectiveness ratios of the various strategies. The probability that LTG first-line therapy is the most cost-effective option remains small, even defining a high cost-effectiveness threshold. Nevertheless, LTG second-line strategies can be cost-effective depending on the willingness to pay for patient improvement. CONCLUSIONS: Only a few studies satisfied our inclusion criteria for employment in our decision model. Our model supports the use of conventional AEDs as first-line options for patients with newly diagnosed epilepsy. LTG second-line therapy is likely to be the most cost-effective option in case society is willing to pay more than Euro 6000 for an additional successfully treated patient. This study also illustrates that, with the data presently available, the outcome of decision analysis for AED treatment choice depends on the inclusion criteria used to select trials. Prospective real-life studies are needed in which first- and second-line treatment strategies are compared with respect to both effectiveness and costs.

Decisions about which health-care interventions represent adequate value to collectively funded health-care systems are as widespread as they are unavoidable. In the case of new pharmaceuticals, many countries now require formal cost-effectiveness analysis to inform this decision-making process. This requires evidence on parameters associated with health-related utilities, treatment effects, resource use, and costs, for which data from available regulatory trials are invariably absent or highly uncertain. This uncertainty results from a number of factors including the predominance of intermediate end points in the clinical evidence-base and the limited period of follow-up of patients in clinical studies. Despite these imperfections in the evidence base, decisions about whether new pharmaceuticals are sufficiently cost-effective for reimbursement cannot be side-stepped. Data limitations do, however, require the use of rigorous analytical methods to support decision making. Probabilistic decision models and value of information analysis offer a means of structuring decision problems, synthesizing all available data, characterizing the uncertainty in the decision, quantifying the cost of uncertainty, and establishing the expected value of perfect information. This analytical framework is important because it addresses two fundamental questions about new pharmaceuticals. First, is the product expected to be cost-effective on the basis of existing evidence? Second, is additional research concerning the product itself cost-effective? In addressing these questions, the analytical framework can establish when sufficient evidence exists to sustain a claim for a new pharmaceutical to be cost-effective.

Background: Modelers and reimbursement decision makers could both profit from a more systematic reporting of the efforts to validate health-economic (HE) models. Objectives: Development of a tool to systematically report validation efforts of HE decision models and their outcomes. Methods: A gross list of model validation techniques was collected using a literature review, including sources outside the HE field. A panel then selected the most important items. Based on the Delphi method, the panel members could score items in three e-mail rounds. Participants were HE modelling experts, covering various nationalities and work environments. They could comment on relevance, feasibility and formulation of the items and received feedback on comments from others. This resulted in a draft tool of selected items, which was tested and improved in two further rounds. In addition, the Dutch National Health Care Institute commented on usefulness for decision makers, while a separate group of 50 HE experts could comment during a workshop at ISPOR Montreal 2014. Results: 35 Validation techniques were identified and grouped into four categories: conceptual model validation, computerized model validation, data validation and operational validation. Around 30 HE experts commented in each of the first three Delphi rounds, resulting in a 15 item draft tool. The Dutch health care advisory institute suggested to add one more item. Participants from the ISPOR workshop delivered 19 filled-in questionnaires. A fourth round resulted in 17 responses. This led to a refined version containing 16 items, which is currently sent out for a final, fifth round. Conclusions: When filled out by the modellers, AdVISHE (Assessment of the ValIdation Status of Health- Economic decision models) supports model users in assessing the validation status of a model It will be useful as part of reimbursement dossiers, by providing systematic and transparent insight into the validation efforts performed and their results.

Survival modeling techniques are increasingly being used as part of decision modeling for health economic evaluations. As many models are available, it is imperative for interested readers to know about the steps in selecting and using the most suitable ones. The objective of this paper is to propose a tutorial for the application of appropriate survival modeling techniques to estimate transition probabilities, for use in model-based economic evaluations, in the absence of individual patient data (IPD). An illustration of the use of the tutorial is provided based on the final progression-free survival (PFS) analysis of the BOLERO-2 trial in metastatic breast cancer (mBC). An algorithm was adopted from Guyot and colleagues, and was then run in the statistical package R to reconstruct IPD, based on the final PFS analysis of the BOLERO-2 trial. It should be emphasized that the reconstructed IPD represent an approximation of the original data. Afterwards, we fitted parametric models to the reconstructed IPD in the statistical package Stata. Both statistical and graphical tests were conducted to verify the relative and absolute validity of the findings. Finally, the equations for transition probabilities were derived using the general equation for transition probabilities used in model-based economic evaluations, and the parameters were estimated from fitted distributions. The results of the application of the tutorial suggest that the log-logistic model best fits the reconstructed data from the latest published Kaplan–Meier (KM) curves of the BOLERO-2 trial. Results from the regression analyses were confirmed graphically. An equation for transition probabilities was obtained for each arm of the BOLERO-2 trial. In this paper, a tutorial was proposed and used to estimate the transition probabilities for model-based economic evaluation, based on the results of the final PFS analysis of the BOLERO-2 trial in mBC. The results of our study can serve as a basis for any model (Markov) that needs the parameterization of transition probabilities, and only has summary KM plots available.

OBJECTIVES: We propose to combine the versatility of the analytic hierarchy process (AHP) with the decision-analytic sophistication of health-economic modeling in a new methodology for early technology assessment. As an illustration, we apply this methodology to a new technology to diagnose breast cancer. METHODS: The AHP is a technique for multicriteria analysis, relatively new in the fi eld of technology assessment. It can integrate both quantitative and qualitative criteria in the assessment of alternative technologies. We applied the AHP to prioritize a more versatile set of outcome measures than most Markov models do. These outcome measures include clinical effectiveness and costs, but also weighted estimates of patient comfort and safety. Furthermore, as no clinical data are available for this technology yet, the AHP is applied to predict the performance of the new technology with regard to all these outcome measures. Results of the AHP are subsequently integrated in a Markov model to make an early assessment of the expected incremental cost-effectiveness of alternative technologies. RESULTS: We systematically estimated priors on the clinical effectiveness and wider impacts of the new technology using AHP. In our illustration, AHP estimates for sensitivity and specifi city of the new diagnostic technology were used as probability parameters in the Markov model. Moreover, the prioritized outcome measures including clinical effectiveness (weight = 0.61), patient comfort (weight = 0.09), and safety (weight = 0.30) were integrated into one outcome measure in the Markov model. CONCLUSIONS: Combining AHP and Markov modelling is particularly valuable in early technology assessment when evidence about the effectiveness of health care technology is still limited or missing. Moreover, combining these methods is valuable when decision makers are interested in other patient relevant outcomes measures besides the technology’s clinical effectiveness, and that may not (adequately or explicitly) be captured in mainstream utility measures.

This paper outlines a decision model for evaluating the effect of information regarding toxicity, efficacy, cost, and alternative treatments on anticipated adoption rates of developmental compounds. METHODS: The paper discusses the relationship between multiattribute utility theory and people's preferences for health outcomes. Health outcomes arising from pharmaceutical interventions are viewed as multiattribute commodities. A method for reliable quantification of health-state preferences is described. This procedure requires estimating utility weights from stated-preference (SP) data. By including health cost as an SP attribute, marginal utilities are converted to marginal dollar values to explicitly account for cost in determining market share. Additionally, SP allows for a nonlinear association between willingness to pay and health states reflecting the diminishing marginal utility of health. Furthermore, this technique permits estimating weights for combined health states. A recent SP design for eliciting respondent trade-offs among episode duration, symptom, daily-activity limitations, and cost is presented. Additionally, various utility specifications and simplifying assumptions are described. Finally, a rule for simulating aggregate choice behaviour is presented. The prediction rule employed draws upon random utility theory. This method addresses the probabilistic nature of choices by including a residual term to represent the effect of unobserved factors on utility. A numerical example demonstrates the capability of the decision model. In this example, a general, preference-based form for health-related utility facilitates direct estimation of health attribute utility weights arising from pharmaceutical product choice. This example demonstrates how the model evaluates the impact of cost, outcomes, and competition on anticipated drug adoption rates. CONCLUSION: Integrating stated-preference data, random-utility modeling, and cost-utility analysis provides a reliable way to describe preferences and predict choice behaviour. This capability provides important information for developmental pharmaceutical product marketing decisions.

Oxford University, Oxford, UKThere was a time when a simple dichotomy charac-terized many health economic evaluations. On theone hand there were those economic appraisals thatwere conducted alongside clinical trials and whichcommonly employed statistical methods in so-calledstochastic evaluations. On the other, there was theuse of decision analytic modeling to synthesize datafrom secondary sources in order to estimate cost-effectiveness in a deterministic fashion. Now, how-ever, the distinctions are becoming ever moreblurred. The limitations of single trials as the solevehicle for economic appraisal is widely reported[1] and, in particular, the continued need for mod-eling to adapt trial-based analyses is well under-stood [2]. Furthermore, the use of probabilisticsensitivity analysis to represent uncertainty in mod-eling studies offers the opportunity to make statis-tical statements about the impact of parameteruncertainty for cost-effectiveness estimates fromdeterministic models.The trend toward a more statistical approach tohandling uncertainty in cost-effectiveness modelshas been picked up in the new guidance from theNational Institute for Clinical Excellence (NICE) inthe UK. NICE now advocates the use of a referencecase set of methods for those making submissions toits appraisal process [3]. Included within this refer-ence case is the recommendation to use probabilisticsensitivity analysis to represent parameter uncer-tainty. Journal editors and reviewers too are becom-ing more stringent in their requirements to see theuse of probabilistic methods to characterize param-eter uncertainty. In this issue of Value in Health,Oostenbrink and colleagues [4] present a probabil-istic model that typifies the approach to presentinguncertainty in such models. By characterizing theuncertainty in the input parameters of their modelas probability distributions, the authors are able topropagate that uncertainty through the model usingMonte Carlo simulation, thus generating a jointdistribution in the incremental costs and effectsthat represents the consequences of the input para-meter uncertainty. This is presented on the cost-effectiveness plane and can be summarized usingcost-effectiveness acceptability curves just as anystatistical analysis of costs and effects from a trial-based analysis.Probabilistic methods are not universallyaccepted, however. A common criticism levied atthe use of probabilistic analysis is that the choice ofdistribution is essentially arbitrary and generatesanother aspect of the analysis that must itself besubjected to sensitivity analysis. Although it is truethat the literature contains many examples of prob-abilistic sensitivity analyses that employ arbitrarydistributional forms, this is generally because theauthors have not thought carefully about the uncer-tainty in the parameter they are seeking to capture.Although there are clearly a large number of poten-tial distributions available in the software packagesused to produce cost-effectiveness models, it doesnot follow that they are all candidate distributionsfor every possible parameter. For any given param-eter in a decision model, the choice of distributionto characterize uncertainty in that parameter shouldbe chosen from a small number of candidate distri-butions which will depend on the data from whichthe parameter is estimated and the method of esti-mation. For example, a normal distribution isalways a candidate for any parameter based onexpected values because of the role of the CentralLimit Theorem, which essentially states that thesampling distribution of the mean will be normallydistributed whatever the underlying distribution ofthe data with sufficient sample size. Of course, mod-els will often be employed in situations where sam-ple sizes of data informing parameter estimationcannot be relied on to be sufficient and other distri-butional forms are required. Probability parame