Crucial Statistical Caveats for Percutaneous Valve Trials

Slides:



Advertisements
Similar presentations
Roseann White Clin/Reg Felllow Abbott Vascular – Cardiac Therapies
Advertisements

Standardized Scales.
Alvin Kwan Division of Information & Technology Studies
ODAC May 3, Subgroup Analyses in Clinical Trials Stephen L George, PhD Department of Biostatistics and Bioinformatics Duke University Medical Center.
Estimation and Reporting of Heterogeneity of Treatment Effects in Observational Comparative Effectiveness Research Prepared for: Agency for Healthcare.
Elements of a clinical trial research protocol
Common Problems in Writing Statistical Plan of Clinical Trial Protocol Liying XU CCTER CUHK.
Large Phase 1 Studies with Expansion Cohorts: Clinical, Ethical, Regulatory and Patient Perspectives Accelerating Anticancer Agent Development and Validation.
Validity Lecture Overview Overview of the concept Different types of validity Threats to validity and strategies for handling them Examples of validity.
Accredited Member of the Association of Clinical Research Professionals, USA Tips on clinical trials Maha Al-Farhan B.Sc, M.Phil., M.B.A., D.I.C.
Qualitative Studies: Case Studies. Introduction l In this presentation we will examine the use of case studies in testing research hypotheses: l Validity;
Adverse Events, Unanticipated Problems, Protocol Deviations & other Safety Information Which Form 4 to Use?
DHHS / FDA / CDRH 1 Circulatory Support Devices Panel Tuesday, September 11, 2001 CoSeal® Surgical Sealant P
Research Design. Research is based on Scientific Method Propose a hypothesis that is testable Objective observations are collected Results are analyzed.
Consumer behavior studies1 CONSUMER BEHAVIOR STUDIES STATISTICAL ISSUES Ralph B. D’Agostino, Sr. Boston University Harvard Clinical Research Institute.
1 Statistical Review Dr. Shan Sun-Mitchell. 2 ENT Primary endpoint: Time to treatment failure by day 50 Placebo BDP Patients randomized Number.
Successful Concepts Study Rationale Literature Review Study Design Rationale for Intervention Eligibility Criteria Endpoint Measurement Tools.
DSBS Discussion: Multiple Testing 28 May 2009 Discussion on Multiple Testing Prepared and presented by Lars Endahl.
How To Design a Clinical Trial
UC c EN. Through Medtronic sponsored research, the Transcatheter Aortic Valves clinical portfolio is studying over 11,000 subjects at over 125.
Investigational Devices and Humanitarian Use Devices June 2007.
Session 6: Other Analysis Issues In this session, we consider various analysis issues that occur in practice: Incomplete Data: –Subjects drop-out, do not.
THE ROLE OF SUBGROUPS IN CLINICAL TRIALS Ralph B. D’Agostino, Sr., PhD Boston University September 13, 2005.
Course: Research in Biomedicine and Health III Seminar 5: Critical assessment of evidence.
Circulatory System Devices Panel Questions for Discussion EMBOL·X Aortic Filter October 23, 2002.
Methodological Issues in Implantable Medical Device(IMDs) Studies Abdallah ABOUIHIA Senior Statistician, Medtronic.
Responsibilities of Sponsor, Investigator and Monitor
Methods of Presenting and Interpreting Information Class 9.
Clinical trials for medical devices: FDA and the IDE process
Rachel Neubrander, PhD Division of Cardiovascular Devices
How To Design a Clinical Trial
Sample size calculation Ahmed Hassouna, MD
Writing a sound proposal
DATA COLLECTION METHODS IN NURSING RESEARCH
These slides highlight a presentation at the Late Breaking Trial Session of the American College of Cardiology 52nd Annual Scientific Sessions in Chicago,
Responsibilities of Sponsor, Investigator and Monitor
The Importance of Adequately Powered Studies
CLINICAL PROTOCOL DEVELOPMENT
FDA’s IDE Decisions and Communications
Dr. Amjad El-Shanti MD, PMH,Dr PH University of Palestine 2016
Statistical Core Didactic
Reasonable Assurance of Safety and Effectiveness: An FDA Division of Cardiovascular Devices Perspective Bram Zuckerman, MD, FACC Director, FDA Division.
Statistical Approaches to Support Device Innovation- FDA View
Clinical Studies Continuum
Data Collection Methods
Within Trial Decisions: Unblinding and Termination
Donald E. Cutlip, MD Beth Israel Deaconess Medical Center
Deputy Director, Division of Biostatistics No Conflict of Interest
Understanding Results
Data, conclusions and generalizations
Medical Device Regulatory Essentials: An FDA Division of Cardiovascular Devices Perspective Bram Zuckerman, MD, FACC Director, FDA Division of Cardiovascular.
Regulatory Considerations for Coronary Drug Coated Balloons – FDA View
Erica Takai, PhD for Andrew Farb, M.D.
PMA Analysis of the CREST Trial Approvability of the RX Acculink Carotid Stent System for Revascularization of Carotid Artery Stenosis in Standard Surgical.
Multiple Endpoint Testing in Clinical Trials – Some Issues & Considerations Mohammad Huque, Ph.D. Division of Biometrics III/Office of Biostatistics/OPaSS/CDER/FDA.
Critical Reading of Clinical Study Results
Is a Clinical Trial Right for Me?
Aiying Chen, Scott Patterson, Fabrice Bailleux and Ehab Bassily
Common Problems in Writing Statistical Plan of Clinical Trial Protocol
Statistical considerations for the Nipah virus treatment study
Project Title Subtitle: make sure you specify it is a research project
Translation Pathway for Coronary Stent Development- Clinical Endpoints
Assessment Literacy: Test Purpose and Use
Experimental Design All experiments consist of two basic structures:
Single Pivotal vs Two Replicated Studies. Zijiang Yang
1 Chapter 8: Introduction to Hypothesis Testing. 2 Hypothesis Testing The general goal of a hypothesis test is to rule out chance (sampling error) as.
Regulatory Perspective of the Use of EHRs in RCTs
Is TCAR best under LA or GA
2019 Joint Statistical Meetings at Denver
Presentation transcript:

Crucial Statistical Caveats for Percutaneous Valve Trials Chul Ahn, Ph.D. FDA/CDRH/OSB/DBS 2/22/2010

DISCLOSURES I have no real or apparent conflicts of interest to report.

Outline Multiple secondary endpoints Data poolability - Pooling sites - Pooling feasibility and pivotal studies Today I’d like to talk about some of the key elements/issues we always discuss with sponsors in all of our studies. They are multiple secondary endpoints and data poolability, especially pooling data between sites and between feasibility and pivotal studies. Might consider pooling data between two delivery systems– transfemoral/transapical. 2/22/2010

Potential Endpoints Primary endpoint: mortality or composite of survival, hospitalization and/or NYHA - claim of statistical significance goes into the indications for use Secondary endpoints: hospitalization, NYHA, MACCE, 6MWT, QOL, Procedural & Clinical Success among others - claim of statistical significance appears in labeling Major Adverse Cardiac and Cerebro-vascular Events (MACCE) 2/22/2010

Multiple Secondary Endpoints An overall significance level 0.05 is to be applied to the safety and effectiveness primary endpoints. Only after the primary endpoint demonstrates statistical significance, the pre-specified secondary endpoints can be tested for statistical significance at an additional overall significance level 0.05. Analysis with post-hoc selected secondary endpoints is exploratory, and may not be able to provide valid statistical evidence of safety and effectiveness in the current confirmatory pivotal trial. 2/22/2010

Multiple Secondary Endpoints A claim of statistical significance from the pre-specified secondary endpoints may only be allowed when hypothesis testing was pre-specified, and multiple testing issues were taken into account. Multiple testing may inflate Type I error It may lead to positive findings purely due to chance. With 10 endpoints (when nothing going on with each of these) and testing each at 0.05, the chance of having at least one false positive finding is as large as 40%. 2/22/2010

Strategies for handling multiplicity Bonferroni Test each of k endpoints at the same level 0.05/k e.g. if k =10, test each endpoint at 0.005 May be too conservative! Two sequential test procedures: Hierarchical closed test procedure Holm’s step-down procedure Hierarchical closed test procedure Order K (>1) multiple secondary endpoints based on clinical importance. If the primary endpoint(s) is statistically significant, then the first important secondary endpoint could be tested at a significance level 0.05. If the first secondary endpoint is significant, then the second important secondary endpoint could be tested at a significance level 0.05. The process continues until the first time a statistical significance testing is failed. Then the procedure stops and the remaining secondary endpoints won’t be tested. Claims of statistical significance could appear in the labeling for those secondary endpoints that passed the test, if allowed by clinical judgment. Holm's step-down procedure (A step-down improvement of the classical Bonferroni method) Order observed univariate p values, and the first secondary endpoint could be declared statistically significant, if p(1) < 0.05/k. Given that the first secondary endpoint is significant, then the second secondary endpoint could be declared statistically significant, if p(2) < 0.05/(k-1). 2/22/2010

Data Poolability Data pooling occurs in many different areas: between studies, multi-centers, physicians or operators, different indications, different patient populations, different device or models, US vs. foreign studies, possibly different body locations and so on. It may be necessary to pool study subjects across different groups in order to obtain adequate sample sizes. However, it is important to validate data pooling. Clinical trials are almost always performed with heterogeneous samples of patients (Simon, 1980), and so, Another consideration: If there is a change in protocol during the study, can the patients afterwards be pooled with those before? In surgical and operator-skilled studies, is there evidence of a learning curve? In device trials, clinical outcomes are often influenced by physician skills. 2/22/2010

Validity of Data Pooling Pooling should be checked in terms of all aspects – clinical, engineering, regulatory, and statistical perspectives. Regulatory considerations: Whether pooling is valid is a question of whether the results under pooling will be valid and interpretable, a criterion that applies to any endpoint, primary or secondary. Statistical concerns: If there are critical clinical, engineering or regulatory concerns on pooling, further statistical efforts on poolability assessment adds no value. It is important to validate data pooling. Perhaps, it should be first checked from clinical and engineering perspectives, and then from regulatory perspectives, and lastly from statistical perspectives. Pooling s/b pre-specified and post-hoc pooling s/b avoided; If pre-specified, is there any penalty for the “interim look”? If post-hoc, study results from the pooling are not be able to provide valid scientific evidence. 2/22/2010

Pooling sites Sponsor’s claim: Statistical analysis showed that there were no site effects in the 30-days MAE rates, which justifies pooling of data across institutions. 2/22/2010

Proposed Poolability Analysis The sponsor only showed that the combined event rates are not different across centers. However, they need to show homogeneity across centers regarding treatment difference (the treatment effects are the same across centers). The combined event rates (the event rates with two treatments combined) 2/22/2010

Are treatment effects the same across centers? (Poolability Question) Clearly, there is a difference in the event rates between treatment and control across sites: a site-by-treatment interaction. 2/22/2010

Site-by-Treatment Interaction In device trials, site-to-site variation can be large. - large variation between sites in physician experience and training in using or implanting the device (sites with device inventor usually perform better) - variation between sites in patient population and patient management When site-by-device interactions are large or qualitative, then a scientifically credible explanation is required. 2/22/2010

Qualitative & Quantitative interaction Qualitative interaction: device effect changes sign depending on the site Quantitative interaction: device effect differ in magnitude, but are in the same direction (model dependent: sometimes, possible to remove them by a transformation of the variable) T T C C site 1 site 2 site 1 site 2 Qualitative interaction Quantitative interaction 2/22/2010

What to do if site-by-treatment interaction is significant? If the interaction is significant, it is may be problematic to interpret the treatment effects directly. If there is significant site-by-treatment interaction, pooling data across site is of concern. The treatment effect should be evaluated by individual site, and any anomalies should be investigated. For example, in superiority trial, we should examine whether any reversal of treatment difference (qualitative interaction) exists. 2/22/2010

Pooling feasibility and pivotal studies Pooling should be avoided as they serve different purposes. Feasibility study is to collect information and generate hypothesis to be tested in a confirmatory study. Pivotal study is to confirm findings obtained from the feasibility study. 2/22/2010

Thanks! chul.ahn@fda.hhs.gov 2/22/2010