Markov versus Medical Markov Modeling – Contrasts and Refinements Gordon Hazen February 2012.

Slides:



Advertisements
Similar presentations
Comparing Two Proportions (p1 vs. p2)
Advertisements

No difference between different types of 3 rd line agent Meta analysis. 19 Articles, c 5k patients Cross et al Ann Int Med 2011.
Factored stochastic tree modeling for medical decision making Gordon Hazen Northwestern University Rowland Chang Northwestern University James Pellissier.
TCOM 501: Networking Theory & Fundamentals
1 Cost-effectiveness analysis using Markov modeling Rahul Ganguly Ph.D. November 25 th, 2006 BITS, Pilani.
Nov M. Huang Northwestern Univ. 1 Markov Chain Population Models in Medical Decision Making Gordon Hazen Min Huang Northwestern University.
The role of economic modelling – a brief introduction Francis Ruiz NICE International © NICE 2014.
Introduction to decision modelling Andrew Sutton.
The Importance of Decision Analytic Modelling in Evaluating Health Care Interventions Mark Sculpher Professor of Health Economics Centre for Health Economics.
Obesity and Cancer: The Link. The Panel’s Recommendation.
Main Points to be Covered
EQUILIBRIUM POPULATION METHODS FOR MARKOV MODELS OF HEALTH INTERVENTIONS Gordon Hazen, Ph.D., Min Huang. IEMS Department, Northwestern University Intervention.
Design of Health Technologies lecture 19 John Canny 11/14/05.
Health care decision making Dr. Giampiero Favato presented at the University Program in Health Economics Ragusa, June 2008.
Economic evaluation of outcomes: long term primary and surrogate endpoints Dr. Giampiero Favato presented at the University Program in Health Economics.
Decision Analysis as a Basis for Estimating Cost- Effectiveness: The Experience of the National Institute for Health and Clinical Excellence in the UK.
Measures of disease frequency (I). MEASURES OF DISEASE FREQUENCY Absolute measures of disease frequency: –Incidence –Prevalence –Odds Measures of association:
BIOST 536 Lecture 4 1 Lecture 4 – Logistic regression: estimation and confounding Linear model.
Sample Size Determination
Calcium/Vitamin D-related CV Events Based on Poster 1163 “Risk of Cardiovascular Events with Calcium/Vitamin D – a Re-Analysis of the Women’s Health Initiative”
Probability, Bayes’ Theorem and the Monty Hall Problem
Critical Appraisal of Clinical Practice Guidelines
Lecture 7  Poisson Processes (a reminder)  Some simple facts about Poisson processes  The Birth/Death Processes in General  Differential-Difference.
Medical Statistics (full English class) Ji-Qian Fang School of Public Health Sun Yat-Sen University.
HSTAT1101: 27. oktober 2004 Odd Aalen
The Nature of Disease.
Decision Models Based on Individual Patient and Summary Data Mark Sculpher Neil Hawkins Centre for Health Economics, University of York Workshop: Towards.
These slides were released by the speaker for internal use by Novartis.
Summary of measures of population Health Farid Najafi MD PhD School of Population Health Kermanshah University of Medical Sciences.
Inference for a Single Population Proportion (p).
Modelling the development of, and treatments for, heart disease and stroke. Tushar Chatterjee, Angus Macdonald & Howard Waters Heriot-Watt University,
Design and Analysis of Clinical Study 11. Analysis of Cohort Study Dr. Tuan V. Nguyen Garvan Institute of Medical Research Sydney, Australia.
Estimating Outcomes in Decision Analysis Brian Harris MPP Candidate Goldman School of Public Policy University of California, Berkeley.
RTI Health Solutions Research Triangle Park North Carolina, US US Manchester, UK UK 44(0) LEADING RESEARCH… MEASURES.
1 © 2008 Brooks/Cole, a division of Thomson Learning, Inc. Chapter 12 The Analysis of Categorical Data and Goodness-of-Fit Tests.
Flows and Networks Plan for today (lecture 4): Last time / Questions? Output simple queue Tandem network Jackson network: definition Jackson network: equilibrium.
Evidence-Based Medicine Diagnosis Component 2 / Unit 5 1 Health IT Workforce Curriculum Version 1.0 /Fall 2010.
Markov Chains X(t) is a Markov Process if, for arbitrary times t1 < t2 < < tk < tk+1 If X(t) is discrete-valued If X(t) is continuous-valued i.e.
Design and Analysis of Clinical Study 10. Cohort Study Dr. Tuan V. Nguyen Garvan Institute of Medical Research Sydney, Australia.
Osteoporosis: Measuring the Problem
This material was developed by Oregon Health & Science University, funded by the Department of Health and Human Services, Office of the National Coordinator.
Measures of Disease Frequency
Satistics 2621 Statistics 262: Intermediate Biostatistics Jonathan Taylor and Kristin Cobb April 20, 2004: Introduction to Survival Analysis.
Conceptual Addition of Adherence to a Markov Model In the adherence-naïve model, medication adherence and associated effectiveness assumed to be trial.
Brennan Spiegel, MD, MSHS
EBM --- Journal Reading Presenter :葉麗雯 Date : 2005/10/27.
Introduction to decision analysis modeling Alice Zwerling, Postdoctoral fellow, JHSPH McGill TB Research Methods Course July 7, 2015.
1 Risk Benefit and Conclusions George Sledge, MD Indiana University School of Medicine.
1 Cost efficacy of smoking cessation interventions Robert West University College London Logroño, October
CT Screening for Lung Cancer vs. Smoking Cessation: A Cost-Effectiveness Analysis Pamela M. McMahon, PhD; Chung Yin Kong, PhD; Bruce E. Johnson; Milton.
Date of download: 5/28/2016 Copyright © 2016 American Medical Association. All rights reserved. From: Practical Issues in Counseling Healthy Women About.
Peripheral Artery Disease in Orthopaedic Patients with Asymptomatic Popliteal Artery Calcification on Plain X-ray Adam Podet, MS; Julia Volaufova, phD,;
Date of download: 5/31/2016 From: Tipping the Balance of Benefits and Harms to Favor Screening Mammography Starting at Age 40 Years: A Comparative Modeling.
Women reaching equality in health, despite the differences Nowadays women remain the largest CAD population subgroup being under-diagnosed and under-treated.
Date of download: 6/21/2016 From: Aspirin, Statins, or Both Drugs for the Primary Prevention of Coronary Heart Disease Events in Men: A Cost–Utility Analysis.
Date of download: 6/24/2016 Copyright © The American College of Cardiology. All rights reserved. From: The Year in Cardiovascular Surgery J Am Coll Cardiol.
Date of download: 6/26/2016 From: Cost-Effectiveness of Adding Cardiac Resynchronization Therapy to an Implantable Cardioverter- Defibrillator Among Patients.
Date of download: 6/26/2016 Copyright © The American College of Cardiology. All rights reserved. From: The Cost-Effectiveness of C-Reactive Protein Testing.
Date of download: 6/27/2016 From: Population Strategies to Decrease Sodium Intake and the Burden of Cardiovascular Disease: A Cost- Effectiveness Analysis.
Date of download: 7/10/2016 From: Potential Cost-Effectiveness of Prophylactic Use of the Implantable Cardioverter Defibrillator or Amiodarone after Myocardial.
How Do We Individualize Guidelines in an Era of Personalized Medicine? Douglas K. Owens, MD, MS VA Palo Alto Health Care System Stanford University, Stanford.
Inference for a Single Population Proportion (p)
Harvard T.H. Chan School of Public Health
The Rise and Fall of Hormone Replacement Therapy
Markov model structure
From: Tipping the Balance of Benefits and Harms to Favor Screening Mammography Starting at Age 40 YearsA Comparative Modeling Study of Risk Ann Intern.
Chapter 8: Inference for Proportions
Health care decision making
No Financial Disclosure or Conflict of Interest
Alcohol, Other Drugs, and Health: Current Evidence
Presentation transcript:

Markov versus Medical Markov Modeling – Contrasts and Refinements Gordon Hazen February 2012

Medical Markov Modeling We think of Markov chain models as the province of operations research analysts However … The number of publications in medical journals – using Markov models – to address medical cost-effectiveness – approaches 300 per year! 2

Medical Markov Modeling Why the large buy-in from the medical community? – Easy-to-use software that combines decision trees and Markov models (Data, TreeAge) – Simplicity of models Discrete time Transient 3

Overview of this talk 1.Background on medical Markov modeling 2.Population modeling versus individual-level modeling 3.Product structure in medical Markov models

Overview 1.Background on medical Markov modeling 2.Population modeling versus individual-level modeling 3.Product structure in medical Markov models.

Medical Markov Modeling The kind of modeling that is typical 6 IHD = Ischemic heart disease MI = Myocardial infarction (heart attack) A simplification of: Palmer S, Sculpher M, Phillips Z, Robinson M, Ginnelly L, Bakhai A et al. Management of non-ST elevation acute coronary syndrome: how cost-effective are glycoprotein IIb/IIIa antagonists in the U.K. National Health Service?. International J Cardiology 100 (2005)

The kind of modeling that is typical Cohort analysis 7

Our preference: Continuous-time Cohort analysis in continuous time 8 p MI =  t p 0 =    t p 1 =    t

Our preference – continuous time Discounted expected quality-adjusted life years: 9

Cohort analysis in continuous time Intervention: Post-MI mortality rate  1 = 0.1/yr is decreased by 75% and the MI incidence rate = 0.12/yr is decreased by 50% QALY/patient10.28 QALY/patient

Continuous-time version of cohort analysis Let dt  0 to obtain 11 … the Kolmogorov differential equations. The cohort analysis procedure is merely the Euler method for solving the Kolmogorov equations.

Overview 1.Background on medical Markov modeling 2.Population modeling versus individual-level modeling 3.Product structure in medical Markov models.

Question: How to incorporate population issues? An intuitive approach: Restart following death 13 Then compute steady-state probabilities in the resulting irreducible chain. Open routing process Closed routing process

Question: How to incorporate population issues? Balance equations for steady-state probabilities: 14 Intervention assumptions: – Post-MI mortality rate  1 = 0.1/yr is decreased by 75% – MI incidence rate = 0.12/yr is decreased by 50%. Results:  PostMI increases from 23.0% to 38.5% The population is less healthy! So what is wrong here?

Population issues: A more rigorous approach Observation: A population of non-interacting individuals is equivalent to a Jackson network of infinite-server queues. 15

Equilibrium results Jackson network balance equations 16 Theorem (e.g. Serfozo 1999): The counts n j of individuals in health state j are, at equilibrium, independent Poisson variables with means  j given by the solution to the balance equations.

Equilibrium results Solve balance equations with entrance rate = 1000/yr More survivors under intervention!

The closed routing process again Convert the open routing process to a closed one in the following way 18 OpenClosed

Open versus closed routing Theorem (Hazen and Huang 2011): One may obtain equilibrium means from steady state probabilities, and vice versa: 19 Equilibrium means  j Steady-state probabilities  j.

Open versus closed results 20 Equilibrium means  j Steady-state probabilities  j.

Example: Re-analysis of the preventive use of tamoxifen Original analysis: Col et al 2002 Tamoxifen – an estrogen agonist/antagonist – an effective therapy against established breast cancer Evidence that it can reduce breast cancer incidence But life-threatening side effects – endometrial cancer – vascular events. Would the benefit of its prophylactic use in healthy women be worth the associated risks? 21

Preventive use of tamoxifen: Our model Cure rate models for breast and endometrial cancer treatment – Mortality decreases in time survived after cancer diagnosis. – This cannot be directly modeled as a Markov model – it is semi-Markov. – Cure rate model with unobserved states Cured/ Not Cured allows implicit mortality to decrease over time survived. 22 Breast cancer incidence and treatment Endometrial cancer incidence and treatment

Preventive use of tamoxifen: Our model Overall model is the Cartesian product of the two factors below and a third Background Mortality factor. More on this later … 23 Breast cancer incidence and treatment Endometrial cancer incidence and treatment

Preventive use of tamoxifen: Our model Estimated parameters (max likelihood estimates) 24

Preventive use of tamoxifen: Our model Product structure for quality of life – Q jk = Q bj Q ek – more on this later 25 Model entry rate 0 = 110,000/yr – 2.3 M women reaching age 50 each year x 4.8% at high risk for breast cancer

Preventive use of tamoxifen: Results A more nuanced picture of the effects of this intervention than just incremental QALYs. 26 Incremental QALYs/woman Incremental equilibrium means Incremental equilibrium probabilities

Overview 1.Background on medical Markov modeling 2.Population modeling versus individual-level modeling 3.Product structure in medical Markov models.

Markov models with product structure Product structure is relatively common in medical Markov models 28 Roach PJ, Fleming C, Hagen MD, Pauker SG. Prostatic cancer in a patient with asymptomatic HIV infection: are some lives more equal than others? Med Decis Making Apr- Jun;8(2):

Markov models with product structure Much simpler depiction of model structure: Independent factors 29

Product structure is relatively common Schousboe et al. considered 5 different types of fractures: – hip fracture – clinical vertebral (Cv) fracture – radiographic vertebral (Rv) fracture – distal forearm (Df) fracture – other fracture In principle this should allow 2 5 = 32 state combinations corresponding to 5 factors each at 2 possible levels. What their model actually did: 6 states – 5 states corresponding to a single fracture type – 1 other state corresponding to the combination of the worst two possible fracture types – Disadvantage: Such a model “forgets” past fractures when a new fracture occurs, which the 32-state model would not do. 30 Schousboe JT, Nyman JA, Kane RL, Ensrud KE. Cost-effectiveness of alendronate therapy for osteopenic postmenopausal women. Ann Intern Med May 3;142(9):

Advantages of explicitly accounting for product structure Model formulation: Simpler to merely consider one factor at a time Model presentation: Simple factors easier to understand and critique. – Model is less likely to be perceived as a “black box” Computational advantages as well when factors are independent. 31

Computation of QALYs under product structure 32

Computation of QALYs under product structure 33

Computation of QALYs under product structure 34

Computation of expected cost under product structure 35

Advantages of factored computation Computational work in cohort analysis is proportional to the number of state transitions Suppose the number of transitions in a factor with s non- death states is roughly s also. Then assuming s states in each factor and f factors, – s f transitions in the overall model under naïve cohort analysis – s  f transitions in cohort decomposition – Big advantage for large f Caveat: s and f are not usually large. 36

Cohort decomposition issues How often are factors independent? – Ans: More often not probabilistically independent. – But one factor is almost always probabilistically independent: Background mortality. How reasonable is the product form for the quality coefficient v(x)? – Empirical support for product form in HUI literature – additive decomposition is not supported. – Often only one factor carries quality adjustments, in which case product form holds by default. 37

Summary These are just the basics – Population modeling Population model  Jackson network One can get at equilibrium population issues by solving the usual balance equations for steady-state probabilities and scaling them up appropriately. – Product structure Common feature of medical Markov models Recognizing it can assist in model formulation and presentation, as well as computation. – Drawbacks for continuous-time models Medical researchers don’t “get” the models Software not widely available There is more to do here …

Questions?