State of the Science: Methods to Collect and Use Patient Preference Data Overview of methods to collect patient preference information and active research regarding patient collecting patient preferences. For the online webcast: Please submit your questions to the panel via the chat box. The online hosts will be collecting the questions during the session to be brought to the panel moderator during the panel discussion.
Moderator: Telba Irony, Ph.D. Chief, General Surgical Devices Branch CDRH/Office of Surveillance and Biometrics Panel: A. Brett Hauber, Ph.D. Research Triangle Institute (RTI ‐ Health Solutions) Bennett Levitan, M.D., Ph.D. Janssen Research & Development (Johnson and Johnson) Sonal Singh, M.D., M.P.H. Johns Hopkins University Scott Braithwaite, M.D. Society of Medical Decision Making (SMDM)
Methods for Eliciting Benefit-Risk Preference Data Patient Preference Initiative Symposium A. Brett Hauber, Ph.D. Senior Economics and Vice President Health Preference Assessment September 18- -19, 2013 19, 2013 September 18
Benefit-Risk Tradeoff Metrics Risk Patient Benefit-Risk Threshold Net Safety Benefit TREATMENT A Maximum Acceptable Risk Δ Risk Effectiveness Δ Effectiveness Net Minimum Effectiveness Acceptable Source: Hauber et al., Appl Health Econ Source: Hauber et al., Appl Health Econ Benefit Benefit 4 4 Health Policy (2013) Health Policy (2013) 4
Benefit-Risk Preference Elicitation Methods • Weighting methods for single decisions Standard Gamble – Threshold Technique – • Generalized weighting methods Best-Worst Scaling – Discrete-Choice Experiments (DCE) – • Decision support methods Analytic Hierarchy Process (AHP) – Multi-Criteria Decision Analysis (MCDA) – 5 5
Benefit-Risk Preference Elicitation Methods • Weighting methods for single decisions Standard Gamble – Threshold Technique – 6 6
Standard Gamble Perfect Perfect Health Health Probability = p Probability = p A A ? Probability = (1- Probability = (1 -p) p) ? Death Death B B Asthma Asthma U(Asthma) = p U(Asthma) = p· ·U(Perfect Health) + (1 U(Perfect Health) + (1- -p p) ) · ·U(Death) U(Death) U (Asthma) = p p· ·(1) + (1 (1) + (1- -p p) )· ·(0) (0) U (Asthma) = MAR (Death) = 1- -p p MAR (Death) = 1 Source: Bernie O Source: Bernie O’ ’Brian, personal Brian, personal communication communication 7 7
Threshold Technique Less pain Less pain Greater Greater hypertension hypertension risk risk Source: Kopec JA et al., Source: Kopec JA et al., J. Clin Epidemiol (2007) J. Clin Epidemiol (2007) 8 8
Benefit-Risk Preference Elicitation Methods • Generalized weighting methods Best-Worst Scaling – Discrete-Choice Experiments (DCE) – 9 9
Best-Worst Scaling (Object Case) MOST LEAST TREATMENT SIDE EFFECT Important Important (Please click ONE) (Please click ONE) Non-fatal stroke Serious infection requiring hospitalization Nausea and vomiting Hand and foot syndrome Bad liver test 10 10
Discrete-Choice Experiment (DCE) Greater Greater weight loss weight loss Longer duration Longer duration of side effects of side effects Lower Risk Lower Risk 11 11
DCE Question: Renal-Cell Carcinoma Efficacy Mild-to Moderate Side Effects Serious Adverse- Event Risks Source: Mohamed et al., Pharmacoeconomics, 2011 Source: Mohamed et al., Pharmacoeconomics, 2011 12 12
Benefit-Risk Thresholds Lung Damage Liver Damage Source: Johnson et al., Chapter 4. Quantifying Patient Source: Johnson et al., Chapter 4. Quantifying Patient 13 13 Preferences to Inform Benefit Preferences to Inform Benefit- -Risk Evaluations ( Risk Evaluations ( In Press In Press ) )
Benefit-Risk Preference Elicitation Methods • Decision support methods Analytic Hierarchy Process (AHP) – Multi-Criteria Decision Analysis (MCDA) – 14 14
Multi-Criteria Decision Analysis (MCDA) Criterion 1 Criterion 2 Criterion 3 Criterion 4 Criterion 5 Alternative 1 a 1 c 1 a 1 c 2 a 1 c 3 a 1 c 4 a 1 c 5 Alternative 2 a 2 c 1 a 2 c 2 a 2 c 3 a 2 c 4 a 2 c 5 Alternative 3 a 3 c 1 a 3 c 2 a 3 c 3 a 3 c 4 a 3 c 5 Alternative 4 a 4 c 1 a 4 c 2 a 4 c 3 a 4 c 4 a 4 c 5 Alternative 5 a 5 c 1 a 5 c 2 a 5 c 3 a 5 c 4 a 5 c 5 Weights W1 W2 W3 W4 W5 where a i c j is the criteria score (performance) of alternative i on criterion j where a c is the criteria score (performance) of alternative i on criterion j i j = Σ Σ Value for Value for = W j a i c j W a c j i j alternative i alternative i j=1- j=1 -n n 15 15
Analytic Hierarchy Process (AHP) Criteria Weighting: : Criteria Weighting Pairwise comparisons Pairwise comparisons on a 9- on a 9 -point scale point scale Tolerabilit Efficacy y AHP Scoring: AHP Scoring : Risk Efficacy Applying weights to Applying weights to Tolerability Risk criteria values criteria values Source: Dolan JG. The Patient 2012 Source: Dolan JG. The Patient 2012 16 16
Multi-criteria Decision Analysis for Patient Multi-criteria Decision Analysis for Patient Preference Assessm ent Preference Assessm ent Bennett Levitan, MD-PhD Departm ent of Epidem iology Janssen Research and Developm ent, Titusville, NJ FDA CDRH Patient Preference I nitiative W orkshop Septem ber 1 8 , 2 0 1 3
Topics • Multi-criteria Decision Analysis for Patient Preferences • I ndustry Considerations for Eliciting Patient Preferences 18
Multi-criteria Decision Analysis ( MCDA) • A generic approach to aid decision-m aking by • Decomposing a complex problem into pieces • Applying judgment to the pieces • Reassembling them into a coherent whole • Generally conducted by a facilitator w ith a sm all group of experts • Long and successful history in Decision Analysis • Lim ited application to benefit-risk or patient considerations until recently • Dodgson JS, Spackman M, Pearman A, Phillips LD. Multi-Criteria Analysis: A Manual. London: Department for Communities and Local Government; 2009 • European Medicines Agency Benefit-Risk Methodology Project, EMA/213482/2010 19
Steps in Multi-criteria Decision Analysis ( MCDA) Clinical Treatments Endpoints Data Data Tradeoffs Context Uncertainty Uncertainty Risk Tolerance Risk Tolerance Linked Decisions Linked Decisions PrOACT URL Decision-making framework •Hammond, Keeney, Raiffa. Smart Choices: A Practical Guide to making Better Decisions. Harvard Bus Sch Press; 1999 •Hunink, Glasziou, et al. Decision Making in Health and Medicine. Cambridge University Press; 2001 20 20
Steps in Multi-criteria Decision Analysis ( MCDA) Clinical Treatments Endpoints Data Data Tradeoffs Context Uncertainty Risk Tolerance Linked Decisions What is the problem I am trying to solve? • • What is the disease and who has it? What is the disease and who has it? • • What is it like to have the disease? What is it like to have the disease? • • How well do existing treatments work? How well do existing treatments work? • • What treatments are we considering? What treatments are we considering? • • Over what time period am I considering treatment? Over what time period am I considering treatment? • • Who is seeing this analysis? Who is seeing this analysis? 21 21
Steps in Multi-criteria Decision Analysis ( MCDA) Clinical Treatments Endpoints Data Data Tradeoffs Context Uncertainty Risk Tolerance Linked Decisions How do I want to characterize treatment performance? Benefit 1 Benefit 1 Benefits Benefits Benefits Benefit 2 Benefits Benefit 2 Benefit 3 Benefit 3 B-R B-R B-R B-R Balance Balance Balance Balance Harm 1 Harm 1 Harm 2 Harm 2 Harms Harms Harms Harms Harm 3 Harm 3 Harm 4 Harm 4 Identify and pare down the key endpoints (attributes) 22 22
Steps in Multi-criteria Decision Analysis ( MCDA) Clinical Treatments Endpoints Data Data Tradeoffs Context Uncertainty Risk Tolerance Linked Decisions How do I want to characterize treatment performance? • • Critical step for benefit-risk Critical step for benefit-risk • • Organizes endpoints into an interpretable tree Organizes endpoints into an interpretable tree • • Distinguishes between critical and less important endpoints Distinguishes between critical and less important endpoints • • Can find differences between what patients consider Can find differences between what patients consider important and what is measured in clinical studies important and what is measured in clinical studies • • Enables identifying symptoms or functional outcomes that Enables identifying symptoms or functional outcomes that can be used for PRO development can be used for PRO development 23 23
Recommend
More recommend
