A RECIPE FOR INCOHERENCE: AVERAGING TIME-TRADEOFF OR STANDARD-GAMBLE UTILITIES ACROSS HEALTH ATTRIBUTES Gordon B. Hazen, IEMS Department, Northwestern.

Slides:



Advertisements
Similar presentations
Elicitation methods Health care demands exceed resource supply Therefore, rationing is inevitable Many ways by which we can ration health care One is economic.
Advertisements

Standardized Scales.
Chapter 8 Flashcards.
VALIDITY AND RELIABILITY
Reliability for Teachers Kansas State Department of Education ASSESSMENT LITERACY PROJECT1 Reliability = Consistency.
Scaling Session Measurement implies “assigning numbers to objects or events…” Distinguish two levels: we can assign numbers to the response levels for.
BAYESIAN POSTERIOR DISTRIBUTIONS FOR PROBABILISTIC SENSITIVITY ANALYSIS Gordon B. Hazen and Min Huang, IEMS Department, Northwestern University, Evanston.
RATING SCALES By M. MUTHU SASI REKHA. SYNOPSIS  Introduction  Definitions  Rating  Principles of defining rating scale  Types of rating scales 
Summary Measures of Population Health: Measuring the impact of disease, injuries and risk factors.
Utility Assessment HINF Medical Methodologies Session 4.
A METHODOLOGY FOR MEASURING THE COST- UTILITY OF EARLY CHILDHOOD DEVELOPMENTAL INTERVENTIONS Quality of improved life opportunities (QILO)
Common Factor Analysis “World View” of PC vs. CF Choosing between PC and CF PAF -- most common kind of CF Communality & Communality Estimation Common Factor.
Introduction to Decision Analysis
Concept of Measurement
COST–EFFECTIVENESS ANALYSIS AND COST-UTILITY ANALYSIS
Chapter 9 Flashcards. measurement method that uses uniform procedures to collect, score, interpret, and report numerical results; usually has norms and.
1 College of Engineering and Computer Science Computer Science Department CSC 131 Computer Software Engineering Fall 2006 Lecture # 2 Chapter 6 & 7 System.
McGraw-Hill © 2006 The McGraw-Hill Companies, Inc. All rights reserved. Correlational Research Chapter Fifteen.
Multivariate Methods EPSY 5245 Michael C. Rodriguez.
Measuring and valuing health outcome Montarat Thavorncharoensap, Ph.D. 1: Faculty of Pharmacy, Mahidol University 2. HITAP, Thailand.
Measurement and Data Quality
1 EQ-5D, HUI and SF-36 Of the shelf instruments…..
1 The valuation of disease-specific questionnaires for QALY analysis  To rescue data in absence of an utility measure  Growth hormone deficiency in adults.
© 2012 Cengage Learning. All Rights Reserved. May not scanned, copied or duplicated, or posted to a publicly accessible website, in whole or in part. Chapter.
Measuring Health Outcomes
Why use the EQ-5D? What are the alternatives?. What are the alternatives for Direct valuation? Other VAS Time Trade-Off Standard Gamble Willingness to.
Some Background Assumptions Markowitz Portfolio Theory
Health Status Adjustment to Initial Barrier-Free Demand Estimate.
Chap 20-1 Statistics for Business and Economics, 6e © 2007 Pearson Education, Inc. Chapter 20 Sampling: Additional Topics in Sampling Statistics for Business.
Health State Unable to perform some tasks at home and/or at work Able to perform all self care activities (eating, bathing, dressing) albeit with some.
University of Minnesota Medical Technology Evaluation and Market Research Department of Healthcare Management Course: MILI/PUBH 6589 Spring Semester, 2013.
Lecture 8: Generalized Linear Models for Longitudinal Data.
Economic evaluation Definition - the comparative analysis of alternative courses of action in terms of both their cost and consequences.
Copyright © Cengage Learning. All rights reserved. Elementary Probability Theory 5.
A Linear Index for Predicting Joint Health States Utilities from Single Health States Utilities Anirban Basu, University of Chicago William Dale, University.
Estimating Outcomes in Decision Analysis Brian Harris MPP Candidate Goldman School of Public Policy University of California, Berkeley.
UNDERSTANDING RESEARCH RESULTS: DESCRIPTION AND CORRELATION © 2012 The McGraw-Hill Companies, Inc.
Correlational Research Chapter Fifteen Bring Schraw et al.
1 The valuation of disease-specific health states to facilitate economic evaluation E. Kok, E. Stolk, Jan J. v. Busschbach Address: –Jan v. Busschbach.
The McGraw-Hill Companies, Inc. 2006McGraw-Hill/Irwin 12 Financial and Cost- Volume-Profit Models.
Investment Analysis and Portfolio Management First Canadian Edition By Reilly, Brown, Hedges, Chang 6.
1 EQ-5D, HUI and SF-36 Of the shelf instruments…..
Evaluating Impacts of MSP Grants Hilary Rhodes, PhD Ellen Bobronnikov February 22, 2010 Common Issues and Recommendations.
Sample Size Determination in Studies Where Health State Utility Assessments Are Compared Across Groups & Time Barbara H Hanusa 1,2 Christopher R H Hanusa.
Introduction to decision analysis Jouni Tuomisto THL.
Chapter 14 Compensation Control & Administration Harcourt, Inc. items and derived items copyright © 2001 by Harcourt, Inc. All rights reserved. Requests.
Chapter Eight: Using Statistics to Answer Questions.
IMPORTANCE OF STATISTICS MR.CHITHRAVEL.V ASST.PROFESSOR ACN.
EBM --- Journal Reading Presenter :呂宥達 Date : 2005/10/27.
Scaling Session Measurement implies assigning numbers to objects or events. In our case, the numbers “weight” responses to questions, so that saying “Yes”
CSCI 347, Data Mining Evaluation: Cross Validation, Holdout, Leave-One-Out Cross Validation and Bootstrapping, Sections 5.3 & 5.4, pages
Chapter 7 Measuring of data Reliability of measuring instruments The reliability* of instrument is the consistency with which it measures the target attribute.
RESEARCH METHODS IN INDUSTRIAL PSYCHOLOGY & ORGANIZATION Pertemuan Matakuliah: D Sosiologi dan Psikologi Industri Tahun: Sep-2009.
Values Lower Than Death Jan J. v. Busschbach, Ph.D. –Erasmus University Rotterdam institute for Medical Technology Assessment (iMTA) PO box DR.
Measurement Experiment - effect of IV on DV. Independent Variable (2 or more levels) MANIPULATED a) situational - features in the environment b) task.
1 Chi-square Test Dr. T. T. Kachwala. Using the Chi-Square Test 2 The following are the two Applications: 1. Chi square as a test of Independence 2.Chi.
Copyright © 2014 Wolters Kluwer Health | Lippincott Williams & Wilkins Chapter 11 Measurement and Data Quality.
2 NURS/HSCI 597 NURSING RESEARCH & DATA ANALYSIS GEORGE MASON UNIVERSITY.
Preference Assessment 1 Measuring Utilities Directly
4 Elementary Probability Theory
Unit 5: Hypothesis Testing
Parts of a Lab Write-up.
Lecture 4: Meta-analysis
EPSY 5245 EPSY 5245 Michael C. Rodriguez
Significance Tests: The Basics
4 Elementary Probability Theory
Chapter Nine: Using Statistics to Answer Questions
Measuring outcomes Emma Frew October 2012.
Elicitation methods Health care demands exceed resource supply
MGS 3100 Business Analysis Regression Feb 18, 2016
Presentation transcript:

A RECIPE FOR INCOHERENCE: AVERAGING TIME-TRADEOFF OR STANDARD-GAMBLE UTILITIES ACROSS HEALTH ATTRIBUTES Gordon B. Hazen, IEMS Department, Northwestern University, Evanston IL Abstract Purpose: Health outcomes are often specified using multiple health attributes. Procedures for assigning QALY coefficients to multiattribute health states include, for example, the Health Utilities Index and the EuroQol. For some cost- effectiveness analyses, the HUI or EuroQol attributes are not specific enough to address important issues. In such cases, modelers may be tempted to assess time-tradeoff or standard gamble utilities one attribute at a time, and then combine the assessed utilities by averaging over attributes. We point out why this procedure is mathematically incoherent, and show what errors in the inferred QALY coefficients may occur as a result. Methods: We consider the case in which health status q = (q 1,q 2 ) is described by two health attributes, and modelers wish to use TTO or SG techniques to assess utilities u 1, u 2, and then form a weighted average to obtain overall QALY coefficient U Q (q) = k 1 u 1 +k 2 u 2, where k 1 and k 2 = 1  k 1 are the weights. We assume that when a subject specifies a TTO or SG response r 1 for a level q 1 of one attribute, s/he by default assumes the other attribute q 2 is at its best level, and vice-versa. Results: Under the averaging model U Q (q) = k 1 u 1 +k 2 u 2, the standard of taking u 1 = r 1 and u 2 = r 2 is no longer valid. If the modeler does so and then averages as just described to obtain the QALY coefficient U Q (q), the resulting theoretical error in U Q (q) is  U Q (q) = (1  r 1 )k 2 + (1  r 2 )k 1. The error  U Q (q) is largest for attribute levels q 1,q 2 farthest below their best possible levels, and can be as large as 0.5 on a scale from 0 to 1 when attributes are equally weighted. The only way to avoid such errors is to replace the averaging rule by the multiplicative combination rule U Q (q) = u 1 u 2. Conclusions: Assessing time-tradeoff or standard gamble utilities one attribute at a time, and then averaging the assessed utilities to obtain an overall QALY coefficient is mathematically incoherent and can lead to large errors in the resulting QALY coefficients. QALY model Standard gamble assessment To obtain the utility U Q (q) of health state q, a subject indicates what chance 1-p of immediate death is worth a p chance at improving health quality from q to q*. The response p is the utility of health state q. Time-tradeoff assessment To obtain the utility U Q (q) of health state q, a subject indicates what reduction in lifetime t 0 would be worth taking to increase health quality to full health q*. The ratio t/t 0 of the response t to the base lifetime t 0 is the utility of health state q. Additively separable QALY model Standard gamble assessment one attribute at a time To obtain the utility u 1 of health state q 1, a subject indicates what chance 1-p of immediate death is worth a p chance at improving health quality from q 1 to q 1 * when health attribute 2 is fixed at some level q 2. The level of q 2 may not be identified, in which case the subject may implicitly assume that q 2 = q 2 *. If the response is p, then the utility u 1 of q 1 may be derived as: Note: The inferred utility u 1 in general is not u 1 = p as it is in the single-attribute case, but rather should depend on the utility u 2 of the second attribute. Taking u 1 = p overestimates u 1 by an amount equal to Time-tradeoff assessment one attribute at at time To obtain the utility u 1 of health state q 1, a subject indicates what reduction in lifetime t 0 would be worth taking to increase health quality to full health q 1 * when health attribute 2 is fixed at some level q 2. Again, the level of q 2 may not be identified, in which case the subject may implicitly assume that q 2 = q 2 *. If the response is t < t 0, then the utility u 1 for state q 1 may be derived as: Note: The inferred utility u 1 in general is again not u 1 = t/t 0 as it is in the single-attribute case, but rather should depend on the utility u 2 of the second attribute. Taking u 1 = t/t 0 overestimates u 1 by an amount equal to Multiplicative QALY model Standard gamble assessment one attribute at a time Note: The inferred utility is u 1 = p, just as it is in the single- attribute case. Time-tradeoff assessment one attribute at at time Note: The inferred utility is u 1 = t/t 0, just as it is in the single-attribute case. Conclusion: Using standard gamble or time tradeoff assessments one attribute at a time 1.Is inconsistent with assuming that health state utility is additively separable across health attributes; 2.Is consistent with assuming that health state utility is multiplicatively separable across attributes. 3.In fact, a multiplicatively separable health state utility function is the only utility function that allows standard gamble or time tradeoffs to be done one attribute at a time (Hazen 2004). Reference GB Hazen (2004), “Multiattribute Structure for QALYs”, forthcoming in Decision Analysis. The total error can be as large as 0.5 on a scale from 0 to 1 when k 1 = k 2 = ½. If both attributes are assessed this way with the other attribute fixed at its best level, the overestimate in U Q is given by The total error can be as large as 0.5 on a scale from 0 to 1 when k 1 = k 2 = ½. If both attributes are assessed this way with the other attribute fixed at its best level, the overestimate in U Q is given by