Course CHL5227H: Introduction to Statistical Methods for Clinical Trials (Fall Semester) Thursdays 9-12am, room HA 316 (Haultain Building) Course Coordinator/instructor: Olli Saarela (olli.saarela@utoronto.ca) Co-instructors: Kevin Thorpe, Janet Raboud, Amy Liu Guest lecturer: Greg Pond Course Overview The aim of this course is to get familiar with the basic concepts and methods in statistical planning, analysis and reporting of evidence in randomized controlled clinical trials. The covered topics will mainly follow the textbook Friedman LM, Furberg CD, DeMets DL (2010): Fundamentals of Clinical Trials, 4 th edition, available as an ebook through U of T library at http://link.springer.com.myaccess.library.utoronto.ca/book/10.1007%2f978-1-4419-1586-3. Additional statistical material is introduced in the lecture materials where necessary. A textbook used as a source material for sessions on randomized screening trials and meta-analysis is Fletcher RH, Fletcher SW, Fletcher GS (2014): Clinical epidemiology: the essentials, 5 th edition, available at http://books1.scholarsportal.info.myaccess.library.utoronto.ca/viewdoc.html? id=/ebooks/ebooks1/lww/2014-07-29/1/01745970. Prerequisites Prior course(s) in statistics or biostatistics. Instructors permission needed for students outside the biostatistics program. Course Objectives The fundamentals of planning and analysis of clinical trials are an integral part of biostatistician's training. At the end of the course, the students should be able to critically evaluate plans and designs for clinical trials, participate in protocol writing, specify a primary research question/hypothesis that can be statistically tested in a trial, specify appropriate intervention and control groups and outcomes, use available design and analysis options to reduce bias and variability of the results, and determine the size of a trial to obtain appropriate power to detect differences. Format of instruction The weekly 3 hour sessions will be in a lecture format, with relevant background reading material listed for each session, to be studied before the class. Lecture slides will also be made available. In addition, there will be in-class midterm and final exams, and four homework assignments; each assignment can include both theoretical and data-analysis exercises.
Evaluation The final grade will consist of a letter grade, and will be determined by the maximum of the weightings 1 and 2 in the below table. Weighting 1 Weighting 2 4 assignments 20% 20% Midterm exam 40% 0% Final exam 40% 80% The assignments are tentatively handed out on September 15, September 29, October 27 and November 10, and are due two weeks from that date, before the class, unless otherwise specified. Late work will not be marked, with the exception of an advance permission from the instructor.
Course Schedule Week Topic Description Reading(s) We will begin with an overview of the course. To introduce the topic, we contrast clinical and epidemiological research, and experimental and nonexperimental research, followed Course by an introduction into the basic outline; Terms 1 (Sep 15) terms and concepts in clinical and concepts trial planning, such as phases, (2010), Chapter 1. of clinical randomization, blinding, trials. blacebos, effectiveness versus efficacy, intent-to-treat principle, and different study designs. Many of these topics will be revisited later in more detail in subsequent sessions. 2 (Sep 22) /KT Inferential statistics: a review We will review the fundamentals of inferential statistics, including the classical hypothesis testing framework, from the clinical trial perspective, where the evidence from the study may serve directly as input to decision making. The necessity of a well-defined primary question/hypothesis is emphasized. The meaning of the normality assumptions routinely employed in statistics is discussed. - (2010), Chapter 3. - Salsburg D (2005): Hypothesis Testing in the Encyclopedia of Biostatistics. - Friedman LM (2005): Clinical Significance Versus Statistical Significance in the Encyclopedia of Biostatistics. 3 (Sep 29) JR Sample size determinations This sessions reviews the basic sample size determination principles for various trial designs, including continuous, binary, time-to-event, and longitudinal outcomes, as well as cross-over and equivalence/ non-inferiority trials. The objective is to understand the motivation behind sample size determinations, and to be able to use the appropriate tool for a given problem. (2010), Chapter 8.
Week Topic Description Reading(s) This session introduces the various multiplicity problems encountered in clinical trials 4 (Oct 6) (e.g. multiple treatments, Multiplicity KT endpoints, and interim analyses), and statistical approaches for correcting for the resulting multiple testing problems. 5 (Oct 13) 6 (Oct 20) 7 (Oct 27) 7 (Nov 3) JR Midterm review In-class midterm exam Causal models Grant applications; Data safety monitoring boards; Pilot studies Review of homework assignments 1 and 2. Design issues and randomization methods are reviewed, with the objective to recognize the different designs and the factors driving the choice of the design. Closed book exam Causal models based on directed acyclic graphs (DAGs) and potential outcome variables are introduced and discussed in the clinical trial context, with the aim of formulating research questions in explicitly causal terms. As an application of this framework, we will introduce the problem of non-compliance, and the instrumental variable approach for estimating complier effects. This sessions introduces the topics of protocol writing, with focus on CIHR grant applications, and interim reporting in clinical trials. Planning and conducting pilot studies is also discussed. Hughes MD (2005): Multiplicity in Clinical Trials in the Encyclopedia of Biostatistics (2010), Chapters 5 and 6. Angrist JD, Imbens, GW and Rubin DB (1996): Identification of causal effects using instrumental variables. Journal of the American Statistical Association, 91:444-455. (2010), Chapter 2. 9 (Nov 10) /AL Meta-analysis of trial evidence This session introduces the most important tools for summarizing evidence and assessing heterogeneity, including funnel and forest plots, fixed effect and random effect meta-analyses, and meta-regression. Fletcher RH et al. (2014), Chapter 13: Summarizing the Evidence.
Week Topic Description Reading(s) Randomized screening trials are introduced and contrasted to therapeutic trials, with the aim of understanding the particular Introduction to 10 (Nov 17) design and analysis issues early detection AL present when studying the trials impact of early detection, such as rare outcomes and delayed effects, and lead time, length and overdiagnosis biases. 11 (Nov 24) GP 12 (Dec 1) 13 (Dec 8) Phase I/II trials Final exam review session Exam week: final exam Early phase clinical trials are discussed, with the aim to understand the questions studied in such trials and the particular design and statistical issues related to these. Review of homework assignments 3 and 4. Statistical analysis issues are reviewed, with the objective of understanding the available methods and the factors driving the choice of the method. The particular focus of statistical modeling in clinical trials is also discussed, for example, the reasons behind covariate adjustment. Closed book exam Fletcher RH et al. (2014), Chapter 10: Prevention. - Storer BE (2005): Phase I Trials - Simor E & Thall PF (2005): Phase II Trials in Encyclopedia of Biostatistics (2010), Chapters 15 and 17.