Society for Medical Decision Making

SMDM Short Courses

Tuesday, October 13, 2020 

9:00 AM - 12:00 PM ET

SMDM Core Course: Introduction to Cost-Effectiveness Analysis
Course Director: Jeffrey S. Hoch, PhD, Division of Health Policy and Management, Department of Public Health Sciences, University of California - Davis, Sacramento, CA

Cost-effectiveness analysis (CEA) is the most popular type of economic evaluation. CEA produces estimates of the extra cost and patient outcomes when one or more new treatments are compared to usual care. CEA is often required by public payers internationally and is often the subject of presentations, posters and papers in academic settings. This course is part of the SMDM core course curriculum.

SMDM Core Course: Introduction to Shared Decision Making and Patient Decision Aids
Course Directors: Victoria Shaffer, PhD, University of Missouri, Columbia, MO; Ellen A. Lipstein, MD, MPH, Cincinnati Children's Hospital Medical Center, Cincinnati, OH

Shared decision making provides a model for patients and their clinicians to engage in a deliberative, communicative process about health decisions in which there is no clear best option from an evidence standpoint. There is an increasing expectation, both nationally and internationally, that patients will be significant partners in decisions about their health. Therefore, it is critical to better understand the state of the science of shared decision making.

Survival Analysis: Why, What, How?
Course Director: Elisabeth Fenwick, PhD, Pharmerit International, Oxford, United Kingdom
Course Faculty: Nicholas Latimer, PhD, School of Health and Related Research, The University of Sheffield, Sheffield, United Kingdom; Sven Klijn, MSc, Pharmerit International, Rotterdam, Netherlands

Determining the clinical value of a product is a key challenge in health technology assessments (HTAs) and survival benefit can be a major component of the clinical value, especially in fields such as oncology. However, survival benefit is typically established based on trial data with limited follow-up, thus extrapolation to a longer time horizon is often required in order to estimate the full value of the product. This course will provide an introduction to why one would want to conduct survival analyses, what methods are available, and how to use these methods. It will cover how to interpret Kaplan-Meier curves, and how to extrapolate observed survival, or any other time to event outcome of interest, to a time horizon of use for HTAs. Participants will explore standard parametric extrapolation methods as well as receiving a primer on alternative methods that can account for more complex survival data.

Applied Cost-Effectiveness Modeling with R
Course Directors: Devin Incerti, PhD, Genentech, South San Francisco, CA; Jeroen Jansen, PhD, University of California - San Francisco; Precision HEOR, San Francisco, CA

Historically, economic models for cost-effectiveness analyses have been developed with specialized commercial software (such as TreeAge) or more commonly with spreadsheet software (almost always Microsoft Excel). But more recently there has been increasing interest in using R and other programming languages for cost-effectiveness analysis, which can offer advantages regarding the integration of input parameter estimation and model simulation, the evaluation of structural uncertainty, quantification of decision uncertainty, incorporation of patient heterogeneity, and computational efficiency, among others. Programming languages such as R also facilitate reproducibility of model-based cost-effectiveness analysis, which is more relevant than ever given recent calls for increased transparency. While these tools are still relatively new, there is an increased interest in learning opportunities as evidenced by recent tutorials, workshops, and development of open-source software.

Advances in Health Outcomes Measurement: Promis and Propr
Course Director: Janel Hanmer, MD, PhD, The University of Pittsburgh, Pittsburgh, PA
Course Faculty: Amy Cizik, PhD, MPH, University of Utah, Salt Lake City, UT; Nan Rothrock, PhD, Northwestern University, Chicago, IL

PROMIS® (Patient-Reported Outcomes Measurement Information System) is a set of person-centered measures that evaluates and monitors physical, mental, and social health in adults and children. It can be used with the general population and with individuals living with chronic conditions. It relies on modern psychometric measurement theory allowing more flexibly in administration than most other PROs. Initially supported by the US National Institutes of Health, it represents the largest, most psychometrically advanced PRO system and has had significant uptake in the US and abroad. The PROMIS-Preference (PROPr) scoring system provides a utility score for PROMIS measurements, meaning that PROMIS data can be used to perform economic analyses such as cost-utility analyses. The course aims to introduce participants to PROMIS and PROPr, including how PROMIS’s psychometric foundation distinguishes it from other PROs and thus distinguishes PROPr from other health-utility measures.

Cost-Effectiveness Analysis with Probabilistic Graphical Models
Course Director: Francisco Javier Diez, PhD, UNED, Madrid, Spain

Participants in this course will learn how to build Bayesian networks, influence diagrams, decision analysis networks, and the Markovian extensions of these models to solve problems of medical diagnosis and cost-effectiveness analysis using OpenMarkov, an open-source tool.

1:00 PM - 4:00 PM ET

SMDM Core Course: Introduction to the Psychology of Medical Decision Making
Course Faculty: KD Valentine, PhD, Division of General Internal Medicine, Massachusetts General Hospital, Boston, MA; Pete Wegier, PhD, Sinai Health, Toronto, ON, Canada

This course introduces participants to psychological theory and empirical research related to making decisions in health and medicine. The psychology of decision making can be used to understand patient and physician behavior and to design behavioral and environmental interventions to improve diagnoses and optimize decision making.

SMDM Core Course: Introduction to Medical Decision Analysis (Decision-Analytic Modeling)
Course Directors Uwe Siebert, MD, MPH, MSc, ScD, UMIT, Dept. of Public Health, Health Services Research & HTA / ONCOTYROL, Division for HTA / Harvard Univ., Dept. Health Policy & Management, Institute for Technology Assessment and Department of Radiology, Hall i. T., Austria / Boston, USA; Beate Jahn, Ass.-Prof, Dr.rer.soc., Dipl Math oec., UMIT - University for Health Sciences, Medical Informatics and Technology; Dept of Public Health, Health Services Research and HTA / Division of Health Technology Assessment; ONCOTYROL - Center for Personalized Cancer Medicine, Hall in Tirol / Innsbruck, Austria

Medical decision making is an essential part of health care. It involves choosing an action after weighing the risks and benefits of the options available to the individual patient or the patient population. While all decisions in health care are made under conditions of uncertainty, the degree of uncertainty depends on the availability, validity, and generalizability of clinical data. Medical decision analysis (or decision-analytic modeling) is a systematic approach to decision making under uncertainty that is used widely in medical decision making, clinical guideline development, and health technology assessment of preventive, diagnostic or therapeutic procedures. It involves combining evidence for different outcomes and from different sources. Outcome parameters may include disease progression, treatment efficacy/effectiveness, safety, quality of life, and individual patient preferences. Sources may include epidemiological studies on the natural history of the disease, randomized clinical trials, observational studies, pharmacoepidemiologic studies, quality of life surveys, risk attitude studies, and others.

An Introduction to Research Prioritisation and Study Design Using Value of Information Analysis
Course Directors: Jeremy D. Goldhaber-Fiebert, PhD, Center for Health Policy and Center for Primary Care and Outcomes Research, Stanford University, Stanford, CA; Anna Heath, PhD, University of Toronto, Toronto, ON, Canada
Course Faculty: Fernando Alarid-Escudero, PhD, Division of Public Administration, Center for Research and Teaching in Economics (CIDE), Aguascalientes, AG, Mexico; Hawre Jalal, MD, PhD, University of Pittsburgh Graduate School of Public Health, Pittsburgh, PA; Natalia Kunst, MSc, Yale University School of Medicine, New Haven, CT

VOI encompasses a suite of measures that quantify the value of reducing parametric uncertainty within a health economic model. VOI measures can determine whether the current evidence base for a health economic decision model is sufficient to make policy decisions. They can also direct future research by determining the model inputs with the greatest influence on decision uncertainty. In addition to this, VOI measures can determine the optimal design for a research study. Despite this versatility, VOI has rarely been used in practice for research prioritization and study design. This is due to a lack of familiarity, difficulties interpreting these measures, concerns about the assumptions underpinning them and computational complexity. The Collaborative Network for Value of Information (ConVOI) group is an international team of experts in developing and applying cutting edge VOI methods that aim to address these challenges.

Microsimulation Modeling in R 
Course Director: Eline Krijkamp, MSc*, Erasmus MC, department of Epidemiology, Rotterdam, Netherlands
Course Faculty: Petros Pechlivanoglou, PhD, Child Health Evaluative Sciences, The Hospital for Sick Children, Toronto, ON, Canada; Fernando Alarid-Escudero, PhD, Division of Public Administration, Center for Research and Teaching in Economics (CIDE), Aguascalientes, AG, Mexico

Many economic evaluations are conducted using Markov cohort models. However, there are many instances where an individual-level model is necessary to capture the clinical realism required for the question of interest. Microsimulation models involve the stochastic simulation of individuals and allow for much greater flexibility over cohort models. Microsimulation models can capture individual clinical pathways, can incorporate complex relationships between clinical history and future events, and more easily capture the impact of heterogeneity in patient demographics, genetics, and other baseline characteristics. Because of their increased complexity, microsimulation models are often implemented in a programming language. The freely available programming environment R can be used to implement, simulate, and analyze the results of a microsimulation model and has parallel processing capabilities, which can improve computational efficiency. In addition, R can facilitate most parts of an evaluation including data analysis to estimate input parameters values as well as documenting model results.

Methods for Measuring Patient Preferences Information
Course Director: John F P Bridges, PhD, The Ohio State University College of Medicine, Columbus, OH
Course Faculty: Ellen Janssen, PhD, Center for Medical Technology Policy (CMTP), Baltimore, MD; Norah Crossnohere, PhD, Johns Hopkins University Bloomberg School of Public Health, Baltimore, MD

There is a growing emphasis on understanding the priorities and preferences of patients and other stakeholders in health and medicine. Patient preference information (PPI) can inform decision making in diverse healthcare settings (e.g. drug development, regulatory assessment, value/health technology assessment, shared decision making). Initiatives are exploring the use of different patient preference methods to inform decision making throughout the entire product lifecycle. Identifying the appropriate preference method requires a thorough understanding of the research question and study objective. To ensure valid and reliable results, good research practices in the development, conduct, analysis, and interpretation of the methods need to be applied. Expanding upon previous short courses on conjoint analysis and discrete-choice experiments (DCE), we will address a larger range of experimental and non-experimental methods for measuring PPI that have been used in health applications. We will also discuss which method is best for different medical decisions and medical decision makers.

Innovative Approaches for Engaging Patients and Improving Decision Making in the Era of Precision Medicine and COVID19
Course Director: Amalia Issa, PhD, MPH, Personalized Medicine & Targeted Therapeutics; and Univ Science Philadelphia, Philadelphia, PA
Course Faculty: Gillian Bartlett, PhD, McGill University, Montreal, QC, Canada

As healthcare continues to undergo rapid transformation with the emergence of precision medicine (PM), AI and other associated technologies, the landscape of medical decision-making is poised to shift remarkably as well. We need new models and approaches in order to inform and assist all stakeholders. To implement PM, providers may need to rely on a combination of types of evidence and algorithms making it difficult to optimally select among different options. This complexity leads to decision challenges, such as a lack of how to properly evaluate health related quality of life. This course introduces participants to new patient-centric focused tools designed to better engage patients in the new era of precision medicine. We will also initiate a dialogue with participants on how to innovate and design better tools that will allow for improved relationships in medical decision making, particularly with measures taken in healthcare due to COVID19.

Introduction to Systematic Review and Meta-Analysis
Course Director: Lisa M. Hess, PhD, Eli Lilly and Co, Indianapolis, IN

Reading and evaluating a meta-analysis is a skill most should have in the field of evidence-based medicine and medical decision making; however, the ability to do so is strengthened not only by understanding and interpreting checklists, but by demonstrating scientific integrity through the knowledge and experience conducting these studies. There are nuances in the conduct of systematic reviews and meta-analysis that can affect the results and their interpretation. There is a need for more individuals to have not only the basic skills to take on these projects, but awareness of to conduct them with accuracy using strong, reproducible research methods. The key skills to understand and apply include selecting fixed effects or random effects models, assessment of heterogeneity and publication bias and the role of sensitivity analyses. This course will support the furthering of these skills so that attendees are confident in proceeding to develop such studies on their own.

Designing and Implementing DICE Simulations of Decision-Analytic Modeling
Course Director: J. Jaime Caro, MDCM, FACP, FRCPC, Evidera, Waltham, MA
Course Faculty: Jorgen Moller, MSc, Evidera, London, United Kingdom

Participants will learn the concepts that are central to a DICE simulation, their formulation, and how to design, specify, build, and use DICE models in MS Excel® to inform various types of health care decisions. For novice, this course will provide a comprehensive introduction to the technique and its use. From previous courses, we find that most students are capable of building their own model afterwards. For experienced modelers, the course will draw on skills they already have but we will show them how to adapt and apply them to the new modality. For reviewers and others who deal with models but don’t necessarily build them, the course will provide the skills required to review, understand, access and work with these tools.

Part 1: An Introduction to Structural Equation Modeling for Medical Decision Making Research
Course Directors: Douglas D. Gunzler, PhD, Center for Health Care Research & Policy, Case Western Reserve University at The MetroHealth System, Cleveland, OH; Dana L. Alden, PhD, University of Hawaii at Manoa, Honolulu, HI; Qimei Chen, PhD, University of Hawaii, Honolulu, HI; Adam T. Perzynski, PhD, Case Western Reserve University at MetroHealth Medical Center, Cleveland, OH; Joseph J. Sudano, PhD, Case Western Reserve University at The MetroHealth System, Cleveland, OH; Kristen Berg, PhD, Case Western Reserve University at MetroHealth, Cleveland, OH

This short course will make Structural Equation Modeling (SEM) accessible to a wide audience of researchers across many disciplines. SEM is a very general and powerful multivariate technique to link conceptual models, path diagrams, factor analysis and other mathematical models. These techniques allow for 1) the combination of continuous, categorical and latent and observed variables; 2) modeling of causal relationships including multiple direct and indirect effects in a single analysis (mediation analysis); 3) cutting-edge techniques for model selection and comparison; 4) measurement equivalence in scales based on sex, race/ethnicity, socioeconomic status, language and other cross-cultural factors: 5) compact representation of cost-utility problems; 6) dynamic updates to model predictions as new clinical measures are obtained; 7) sophisticated approaches to modeling change over time. These advantages are particularly applicable to both theoretical and applied research problems in medical decision making.

Tuesday, October 20, 2020

9:00 AM - 12:00 PM ET

Causal Inference and Causal Diagrams in Medical Decision Making Using Big Real World Observational Data and Pragmatic Trials
Course Director Uwe Siebert, MD, MPH, MSc, ScD, UMIT, Dept. of Public Health, Health Services Research & HTA / ONCOTYROL, Division for HTA / Harvard Univ., Dept. Health Policy & Management, Institute for Technology Assessment and Department of Radiology, Hall i. T., Austria / Boston, USA
Course Faculty Felicitas Kuehne, MSc, Institute of Public Health, Medical Decision Making and Health Technology Assessment, Department of Public Health, Health Services Research and Health Technology Assessment, UMIT - University for Health Sciences, Medical Informatics and Technology, Hall i.T., Austria

One of the most important tasks of decision makers is to derive causal interpretations using both statistical analyses of original datasets and decision analysis. Often an intervention, action or risk factor is modeled to have a "causal effect" on one or more model parameters (e.g., probability, rate, or mean of outcome). Therefore, both the biostatistician and the decision analyst need tools to check: (1) when effect estimates have a causal interpretation and when they do not; and (2) the appropriate methods to derive causal effects instead of merely statistical associations (e.g., traditional multivariate regression analysis or causal g-methods). This course intends to provide basic knowledge on causal thinking and visual, structural, and statistical tools to be able to judge whether estimates are suitable for causal interpretation. We will also very briefly touch on the topic “Causal Discovery”, that is, deriving the causal diagram from the data.

Advanced Computation of Value of Information Measure to Determine Optimal Research Design
Course Directors: Anna Heath, PhD, University of Toronto, Toronto, ON, Canada; Jeremy D. Goldhaber-Fiebert, PhD, Center for Health Policy and Center for Primary Care and Outcomes Research, Stanford University, Stanford, CA
Course Faculty: Hawre Jalal, MD, PhD, University of Pittsburgh Graduate School of Public Health, Pittsburgh, PA; Fernando Alarid-Escudero, PhD, Division of Public Administration, Center for Research and Teaching in Economics (CIDE), Aguascalientes, AG, Mexico; Natalia Kunst, MSc, Yale University School of Medicine, New Haven, CT

EVSI has long been touted as a method for research prioritisation and study design. Based on a health economic decision model, EVSI computes the monetary value of performing research. This value can be compared with the cost of conducting the research to determine whether there it has a potential positive net benefit. Despite these advantages, the computational complexity of this type of analysis has hindered its implementation in practice. Recently, methods have been developed to reduce this computational burden and allow for the estimation of EVSI in practice. The Collaborative Network for Value of Information (ConVOI) group is a team of researchers, including the developers of four recent EVSI estimation methods, that aims to improve the visibility and implementation of EVSI in research prioritization and study design.

Quantifying and Valuing Health Inequality Impacts in Economic Evaluation
Course Director: Susan Griffin, PhD, Centre for Health Economics, University of York, York, United Kingdom
Course Faculty: Fan Yang, PhD, Centre for Health Economics, University of York, York, United Kingdom

The lack of formal analysis of health inequality impacts in health technology assessment and economic evaluation may hinder their influence on decision making. This course addresses that gap by teaching participants how to quantify health inequality impacts in economic evaluation. This course will teach participants how to: (i) use subgroup analysis within economic evaluation to describe a distribution of health with and without an intervention; (ii) calculate the level of inequality in the distribution of health; (iii) construct a health equity impact plane; (iv) combine total health impact and impact on health inequality into a single summary measure in terms of either societal welfare or health; (v) estimate and interpret equally distributed equivalent health. These skills are used to enable researchers to formally incorporate health inequality impacts into economic evaluation. They are used to estimate and communicate the extent and value of health inequality impacts to stakeholders and decision makers.

Hands-on Model Calibration in R 
Course Director: Eva A. Enns, MS, PhD, University of Minnesota School of Public Health, Minneapolis, MN
Course Faculty: Fernando Alarid-Escudero, PhD, Division of Public Administration, Center for Research and Teaching in Economics (CIDE), Aguascalientes, AG, Mexico

In developing mathematical models of disease processes for medical decision making, there are often a subset of model parameters that cannot be observed for physical, practical, or ethical reasons. For example, cancer progression rates prior to detection cannot, by definition, be directly observed. Calibration is the process by which values of uncertain or unknown parameters are estimated such that model outputs match observed clinical or epidemiological data (“calibration targets”). Generally, calibration involves two main components: 1) a strategy for searching through the space of the unknown parameters; and 2) a goodness-of-fit measure that reflects how well a set of model outputs matches the target data. In this course, we will cover how to implement different approaches to each of these steps in R. We will also provide guidance on the pros and cons of different approaches and the circumstances under which some approaches may be more appropriate than others.

Advancing the Critical Appraisal of Methods to Measure Patient Preferences
Course Director: John F P Bridges, PhD, The Ohio State University College of Medicine, Columbus, OH
Course Faculty: Ellen Janssen, PhD, Center for Medical Technology Policy (CMTP), Baltimore, MD; Sara Knight, PhD, University of Utah, Salt Lake City, UT

The study of patient preferences aims to assess the relative desirability or acceptability of health states, interventions, or programs. Multiple methods are now used to assess patient preferences spanning qualitative and quantitative approaches, experimental and non-experimental methods, and relatively simple to relatively complex studies. There is a growing literature on the development, application, reporting and dissemination, and, most recently, on the use of patient preference data by decision makers in medicine. These approaches have focused on standardizing terminology, methods, and publishing practices, often taking a cook-book approach towards developing better instruments and publishing more complete and transparent papers. Most recently the US FDA has highlighted desirable guiding principles of preference studies, but certain elements of this framework remain difficult to operationalize and assess (e.g. relevance and the study of heterogeneity).

Modeling Approaches for Analyzing Health Care Problems – an Introductory Overview and Comparison
Course Director: Beate Jahn, Ass.-Prof, Dr.rer.soc., Dipl Math oec., UMIT - University for Health Sciences, Medical Informatics and Technology; Dept of Public Health, Health Services Research and HTA / Division of Health Technology Assessment; ONCOTYROL - Center for Personalized Cancer Medicine, Hall in Tirol / Innsbruck, Austria
Course Faculty: Mark S. Roberts, MD, MPH, Health Policy and Management, University of Pittsburgh Graduate School of Public Health, Pittsburgh, PA; James Stahl, MD, CM, MPH, Dartmouth-Hitchcock Medical Center, Lebanon, NH; Uwe Siebert, MD, MPH, MSc, ScD, UMIT, Dept. of Public Health, Health Services Research & HTA / ONCOTYROL, Division for HTA / Harvard Univ., Dept. Health Policy & Management, Institute for Technology Assessment and Department of Radiology, Hall i. T., Austria / Boston, USA

Decision-analytic modeling is increasingly applied to analyze decisions under conditions of uncertainty to allocate limited resources in health care. Decisions range from the evaluation of preventions, diagnostic or treatment up to scheduling and planning of health care resources. Decision-analytic models are powerful tools allowing estimation of long-term benefits, risks and harms. Intermediate outcomes of clinical trials can be linked with long term observational studies and cost-effectiveness studies can be done across jurisdictions. The appropriate model type is determined by the research question, nature of the disease, required level of detail, and complexity. Commonly used modeling techniques are: 1) Decision Trees (DT), 2) State Transition Markov Models (STMM), 3) State Transition Microsimulation Models 4) Discrete-Event-Simulation (DES), 5) Agent-based Models (ABM), and 6) System Dynamics (SD).

Part 2: An Introduction to Structural Equation Modeling for Medical Decision Making Research
Course Directors: Douglas D. Gunzler, PhD, Center for Health Care Research & Policy, Case Western Reserve University at The MetroHealth System, Cleveland, OH; Dana L. Alden, PhD, University of Hawaii at Manoa, Honolulu, HI; Qimei Chen, PhD, University of Hawaii, Honolulu, HI; Adam T. Perzynski, PhD, Case Western Reserve University at MetroHealth Medical Center, Cleveland, OH; Joseph J. Sudano, PhD, Case Western Reserve University at The MetroHealth System, Cleveland, OH; Kristen Berg, PhD, Case Western Reserve University at The MetroHealth System, Center for Health Care Research and Policy, Cleveland, OH

This short course will make Structural Equation Modeling (SEM) accessible to a wide audience of researchers across many disciplines. SEM is a very general and powerful multivariate technique to link conceptual models, path diagrams, factor analysis and other mathematical models. These techniques allow for 1) the combination of continuous, categorical and latent and observed variables; 2) modeling of causal relationships including multiple direct and indirect effects in a single analysis (mediation analysis); 3) cutting-edge techniques for model selection and comparison; 4) measurement equivalence in scales based on sex, race/ethnicity, socioeconomic status, language and other cross-cultural factors: 5) compact representation of cost-utility problems; 6) dynamic updates to model predictions as new clinical measures are obtained; 7) sophisticated approaches to modeling change over time. These advantages are particularly applicable to both theoretical and applied research problems in medical decision making.