Sensitivity Analysis of Randomized Trials with Missing Data Daniel Scharfstein Department of Biostatistics Johns Hopkins University

Slides:



Advertisements
Similar presentations
1 Statistical Analysis for an AIDS Clinical Trial Weiming Ke The University of Memphis St. Jude Childrens Research Hospital June 5, 2004.
Advertisements

1 Contact details Colin Gray Room S16 (occasionally) address: Telephone: (27) 2233 Dont hesitate to get in touch.
Allison Dunning, M.S. Research Biostatistician
New Multiple Sclerosis Diagnostic Criteria CLINICAL ATTACKS OBJECTIVEADDITIONAL REQUIREMENTS 2 or more None; clinical evidence will suffice 2 or more1Dissemination.
T. A. LouisTrialNet Workshop March 7, The POPPI 1 Example: Statistical Comments Thomas A. Louis, PhD Department of Biostatistics Johns Hopkins Bloomberg.
If we use a logistic model, we do not have the problem of suggesting risks greater than 1 or less than 0 for some values of X: E[1{outcome = 1} ] = exp(a+bX)/
Controlling for Time Dependent Confounding Using Marginal Structural Models in the Case of a Continuous Treatment O Wang 1, T McMullan 2 1 Amgen, Thousand.
Grandparenting and health in Europe: a longitudinal analysis Di Gessa G, Glaser K and Tinker A Institute of Gerontology, Department of Social Science,
1 QOL in oncology clinical trials: Now that we have the data what do we do?
Analysis & Expressing Resultd in Clinical Trials Dr. Khalili.
CCEB Modeling Quality of Life Data with Missing Values Andrea B. Troxel, Sc.D. Assistant Professor of Biostatistics Center for Clinical Epidemiology and.
Experimental Design making causal inferences. Causal and Effect The IV precedes the DV in time The IV precedes the DV in time The IV and DV are correlated.
Journal Club Alcohol, Other Drugs, and Health: Current Evidence May–June 2013.
Confidence Intervals © Scott Evans, Ph.D..
Journal Club Alcohol, Other Drugs, and Health: Current Evidence September–October 2008.

Common Problems in Writing Statistical Plan of Clinical Trial Protocol Liying XU CCTER CUHK.
1 Equivalence and Bioequivalence: Frequentist and Bayesian views on sample size Mike Campbell ScHARR CHEBS FOCUS fortnight 1/04/03.
Augmented designs to assess immune responses in vaccine efficacy trials Talk adapted from Dean Follmann’s slides NIAID Biostat 578A Lecture 12.
Raymond J. Carroll Texas A&M University LOCF and MMRM: Thoughts on Comparisons.
Revisiting an Old Topic: Probability of Replication D. Lizotte, E. Laber & S. Murphy Johns Hopkins Biostatistics September 23, 2009.
Journal Club Alcohol, Other Drugs, and Health: Current Evidence November–December 2008.
Today Today: Finish Chapter 9, start Chapter 10 Sections from Chapter 9: 9.1, 9.4, 9.5, 9.10 (know just class notes for these sections) Recommended Questions:
ACTG 333 The Antiviral Effect of Switching from Saquinavir to the New Formulation of Saquinavir vs. Switching to Indinavir After >1 year of Saquinavir.
BC Jung A Brief Introduction to Epidemiology - XI (Epidemiologic Research Designs: Experimental/Interventional Studies) Betty C. Jung, RN, MPH, CHES.
The Mimix Command Reference Based Multiple Imputation For Sensitivity Analysis of Longitudinal Trials with Protocol Deviation Suzie Cro EMERGE.
What is a Clinical Trial (alpha version) John M. Harris Jr., MD President Medical Directions, Inc.
Experimental Design making causal inferences Richard Lambert, Ph.D.
Assessing ETA Violations, and Selecting Attainable/Realistic Parameters Causal Effect/Variable Importance Estimation and the Experimental Treatment Assumption.
Decisions from Data: The Role of Simulation Gail Burrill Gail Burrill Michigan State University
Part III Gathering Data.
Types of Research and Designs This week and next week… Covering –Research classifications –Variables –Steps in Experimental Research –Validity –Research.
Alec Walker September 2014 Core Characteristics of Randomized Clinical Trials.
Abstinence Incentives for Methadone Maintained Stimulant Users: Outcomes for Those Testing Stimulant Positive vs Negative at Study Intake Maxine L. Stitzer.
HIV-1 dynamics Perelson et.al. Science 271:1582 (1996) Infected CD4 + lymphocytes Uninfected, activated CD4 + lymphocytes HIV-1 t 1/ days t 1/2.
Empirical Efficiency Maximization: Locally Efficient Covariate Adjustment in Randomized Experiments Daniel B. Rubin Joint work with Mark J. van der Laan.
MISSING DATA IN THE INFECTIOUS DISEASES INSTITUTE CLINIC DATABASE Agnes N Kiragga East Africa IeDEA investigators’ meeting 4-5 th May 2010 East African.
Biostatistics Case Studies 2006 Peter D. Christenson Biostatistician Session 4: An Alternative to Last-Observation-Carried-Forward:
4.2 Binomial Distributions
The Diabetic Retinopathy Clinical Research Network Effect of Diabetes Education During Retinal Ophthalmology Visits on Diabetes Control (Protocol M) 11.
Simulation Study for Longitudinal Data with Nonignorable Missing Data Rong Liu, PhD Candidate Dr. Ramakrishnan, Advisor Department of Biostatistics Virginia.
A shared random effects transition model for longitudinal count data with informative missingness Jinhui Li Joint work with Yingnian Wu, Xiaowei Yang.
Sensitivity Analysis for Interval-Censored Discrete Failure Time Data: Application to ACTG 181 Daniel Scharfstein Department of Biostatistics Johns Hopkins.
1 Pulminiq™ Cyclosporine Inhalation Solution Pulmonary Drug Advisory Committee Meeting June 6, 2005 Statistical Evaluation Statistical Evaluation Jyoti.
Journal Club Alcohol, Other Drugs, and Health: Current Evidence September–October 2013.
ACTG 5142: First-line Antiretroviral Therapy With Efavirenz Plus NRTIs Has Greater Antiretroviral Activity Than Lopinavir/Ritonavir Plus NRTIs Slideset.
Reference based sensitivity analysis for clinical trials with missing data via multiple imputation Suzie Cro 1,2, Mike Kenward 2, James Carpenter 1,2 1.
Reference based multiple imputation;
Methods to Handle Noncompliance
Sample Size Considerations
Statistical Modelling
The Importance of Adequately Powered Studies
Alcohol, Other Drugs, and Health: Current Evidence
MISSING DATA AND DROPOUT
Lopinavir-ritonavir mg BID (n = 354)
Sensitivity analyses for missing not at random outcomes in clinical trials (Invited Session 1.2: Recent Advances in Methods for Handling Missing Data in.
Aligning Estimands and Estimators – A Case Study Sept 13, 2018 Elena Polverejan Vladimir Dragalin Quantitative Sciences Janssen R&D, Johnson & Johnson.
Alcohol, Other Drugs, and Health: Current Evidence May-June, 2018
Mark Rothmann U.S. Food and Drug Administration September 14, 2018
Aiying Chen, Scott Patterson, Fabrice Bailleux and Ehab Bassily
Common Problems in Writing Statistical Plan of Clinical Trial Protocol
Evaluating research Is this valid research?.
Interpreting Epidemiologic Results.
What Do We Know About Estimators for the Treatment Policy Estimand
HEC508 Applied Epidemiology
Clinical prediction models
Use of Piecewise Weighted Log-Rank Test for Trials with Delayed Effect
Changing Education Paradigms
A Bayesian Design with Conditional Borrowing of Historical Data in a Rare Disease Setting Peng Sun* July 30, 2019 *Joint work with Ming-Hui Chen, Yiwei.
How Should We Select and Define Trial Estimands
Presentation transcript:

Sensitivity Analysis of Randomized Trials with Missing Data Daniel Scharfstein Department of Biostatistics Johns Hopkins University

ACTG 175 ACTG 175 was a randomized, double bind trial designed to evaluate nucleoside monotherapy vs. combination therapy in HIV+ individuals with CD4 counts between 200 and 500. Participants were randomized to one of four treatments: AZT, AZT+ddI, AZT+ddC, ddI CD4 counts were scheduled at baseline, week 8, and then every 12 weeks thereafter. Additional baseline characteristics were also collected.

ACTG 175 One goal of the investigators was to compare the treatment-specific means of CD4 count at week 56 had all subjects remained on their assigned treatment through that week. The interest is efficacy rather than effectiveness. We define a completer to be a subject who stays on therapy and is measured at week 56. Otherwise, the subject is called a drop-out. 33.6% and 26.5% of subjects dropped out in the AZT+ddI and ddI arms, respectively.

ACTG 175 Completers-only analysis Treatment Mean CD4 SE95% CI AZT+ddI ddI Difference2511.5(3,48) p=0.0027

ACTG 175 The completers-only means will be valid estimates if, within treatment groups, the completers and drop-outs are similar on measured and unmeasured characteristics. Missing at random (MAR), with respect to treatment group. Without incorporating additional information, the MAR assumption is untestable. It is well known from other studies that, within treatment groups, drop-outs tend to be very different than completers.

Goal Present a coherent paradigm for the presentation of results of clinical trials in which it is plausible that MAR fails (i.e., NMAR). Sensitivity Analysis Bayesian Analysis

Sensitivity Analysis For each treatment group, specify a set of models for the relationship between the distributions of the outcome for drop-outs and completers. Index the treatment-specific models by an untestable parameter (alpha), where zero denotes MAR. alpha is called a selection bias parameter and it indexes deviations from MAR. Pattern-mixture model Step 1: Models

Treatment-specific imputed distributions of CD4 count at week 56 for drop-outs

Step 1: Models Selection model The parameter alpha is interpreted as the log odds ratio of dropping out when comparing subjects whose log CD4 count at week 56 differs by 1 unit. alpha>0 (<0) indicates that subjects with higher (lower) CD4 counts are more likely to drop-out. alpha=0.5 (-0.5) implies that a 2-fold increase in CD4 count yields a 1.4 increase (0.7 decrease) in the odds of dropping out. Sensitivity Analysis

Step 2: Estimation For a plausible range of alphas, estimate the treatment-specific means by taking a weighted average of the mean outcomes from the completers and drop-outs. Sensitivity Analysis

Treatment-specific imputed distributions of CD4 count at week 56 for drop-outs

Treatment-specific estimated mean CD4 at week 56 as function of alpha

Step 3: Testing Test the null hypothesis of no treatment effect as a function of treatment-specific selection bias parameters. For each combination of the treatment- specific selection bias parameters, form a Z-statistic by taking the difference in the estimated means divided by the estimated standard error of the difference. Sensitivity Analysis

If the selection bias parameters are correctly specified, this statistic is normal(0,1) under the null hypothesis. Reject the null hypothesis at the 0.05 level if the absolute value of the Z-statistic is greater than Step 3: Testing Sensitivity Analysis

Contour Plot of Z-statistic

Bayesian Analysis Think of all model parameters as random. Place prior distributions on these parameters. Informative prior on alpha (e.g., normal with mean -0.5 and standard deviation 0.25). Non-informative priors on all other parameters (e.g., the distribution of the outcome). Results are summarized through posterior distributions.

Posterior Distributions 368 (342,391) 348 (330,365)

Posterior Distribution of Mean Difference 20 (-11,49); 91%

Likelihood-based Inference A parametric model for the outcome and a parametric for the probability of being a completer given the outcome. For example, the outcome is log normal. Inference proceeds by maximum likelihood (ML). ML inference can be well approximated using Bayesian machinery.

Maximum Likelihood Distributions 303 (278,331) 297 (271,324) 368 (342,391) 348 (330,365) -2.6 (-3.0,-2.1) -2.8 (-3.3,-2.2)

Treatment-specific imputed distributions of CD4 count at week 56 for drop-outs

Maximum Likelihood Distribution of Mean Difference 20 (-11,49)7 (-31,44)

Incorporating Auxiliary Information MAR (with respect to all observable data) Sensitivity analysis with respect alpha. Bayesian methods under development. Longitudinal/Time-to-Event Data Same underlying principles.

LOCF Bad idea Imputing an unreasonable value. Results may be conservative or anti- conservative. Uncertainty is under-estimated.

Conjecture There is information from previously conducted clinical studies to help in the analysis of the current trials. Data from previous trials may be able to restrict the range of or estimate alpha.

Summary We have presented a paradigm for reporting the results of clinical trials where missingness is plausibly related to outcomes. We believe this approach provides a more honest characterization of the overall uncertainty, which stems from both sampling variability and lack of knowledge of the missingness mechanism.

Scharfstein, Rotnitzky A, Robins JM, and Scharfstein DO: Semiparametric Regression for Repeated Outcomes with Non-ignorable Non-response, Journal of the American Statistical Association, 93, , Scharfstein DO, Rotnitzky A, and Robins, JM.: Adjusting for Non-ignorable Drop-out Using Semiparametric Non-response Models (with discussion), Journal of the American Statistical Association, 94, , Rotnitzky A, Scharfstein DO, Su TL, and Robins JM: A Sensitivity Analysis Methodology for Randomized Trials with Potentially Non-ignorable Cause-Specific Censoring, Biometrics, 57:30-113, 2001 Scharfstein DO, Robin JM, Eddings W and Rotnitzky A: Inference in Randomized Studies with Informative Censoring and Discrete Time-to- Event Endpoints, Biometrics, 57: , Scharfstein DO and Robins JM: Estimation of the Failure Time Distribution in the Presence of Informative Right Censoring, Biometrika 89: , Scharfstein DO, Daniels M, and Robins JM: Incorporating Prior Beliefs About Selection Bias in the Analysis of Randomized Trials with Missing Data, Biostatistics, 4: , 2003.