Industries: Pharma / BiotechMedical Device

Advanced Topics in Biostatistics for Non-Statisticians™

Course Director: Al Bartolucci, Ph.D.

Course Description - Course runs 9:00 to 5:00 both days

This pharmaceutical and medical device training course begins with a brief introductory discussion that will introduce and outline the types of clinical investigations conducted. This includes Phase II (non- randomized and randomized) and Phase III randomized clinical trials. It will emphasize the principles of clinical investigations and the issues to be addressed in the remainder of the course. Also the concept of the p-value and power will be reviewed.

Statistical topics will include, but are not limited to:

  • Multiple primary and secondary endpoints in clinical trials and the techniques for addressing the multiple testing procedures that have been proposed
  • Missing data in clinical trials, the types of missing data that can arise and the issues involved for handling and interpreting the results from lack of data
  • Adaptive designs in clinical studies when an interim analysis indicates a possible midway correction in study design to preserve the integrity of the study
  • Addressing the problems of assessing risks in time to event studies when failure may be from competing causes. The course will cover more complex issues in research investigations and clinical trials.
  • Introduction to Bayes Analysis in Clinical Studies and their comparison to the traditional hypothesis testing procedures
  • Propensity scoring in the presence of many covariates and the advantages of this technique in many clinical studies
  • Statistical approaches to high density data involving genetic markers as diagnostic or treatment correlates
  • Brief introduction to gaining more information from meta-analysis using meta-regression and other techniques

Who Should Attend

This two-day course is designed as an overview to the statistical principles that go beyond the basics for the design and analysis of research investigations in pharmaceutical and medical device studies. The focus of topics will benefit individuals within the pharmaceutical, biotech and medical device industries including medical investigators, basic and clinical research scientists, clinical research associates those involved in regulatory affairs who have taken the introductory course "Biostatistics for Non–Statisticians" or are familiar with the basic conduct of clinical trials and wish to pursue a deeper understanding of statistical principles.

This biotech training program will concentrate on the philosophy and understanding of the statistical requirements used in conducting sound scientific investigations. It will not simply present statistical formulae. Thus, the lectures are oriented toward professional who have familiarity with basic principles of Biostatistics and/or Statistical Analysis or at least have attended the basic course "Biostatistics for Non-Statisticians".

First Day

Multiple Primary and Secondary Endpoints: Reasons for composite endpoints, types of multiplicity encountered, design issues, examples of statistical procedures for addressing multiple endpoints, weighting of hypotheses, multiple stage testing procedures.

Missing Data in Clinical Trials: Types of missing data, mechanisms of missing data (e.g. missing completely at random, missing at random, non-ignorable), consequences of missing data, simple and multiple imputation examples.

Adaptive Designs: Midway correction techniques, preservation of Type I and Type II errors, two and three stage design examples (randomized and non-randomized), comparison to group sequential (interim analysis only) designs.

Competing Risk Analysis: Definitions of types of competing risks (CR),Types of trials with CR, the different statistical approaches to CR (advantages, disadvantages), several example comparing performance of these approaches.

Second Day

Introduction to Bayes Analysis of Clinical and Epidemiological Studies: Explanation of Bayes methods, Comparison of Bayes methods to traditional statistical approaches to clinical studies such as screening designs, drug development analysis, equivalence studies. Time to event (survival) prediction using Bayes methods. Propensity Scoring in Clinical Trials: Definition and popularity of propensity scoring, the need for propensity measures in place of large sets of covariates, Types of propensity scoring (modeling, matching, stratification) and detailed data examples of each.

Statistical Approaches to High Density Data Sets: Classification techniques according to diagnostic or treatment assignment outcomes, decision trees, bootstrap forest and boost trees, the role of ROC curves, recent methodology for analysis gene expression data.

Brief Introduction to Meta-Regression: Methods of meta-regression beyond the usual meta-analysis to determine variables or covariates influencing outcome other than the main treatment /intervention effect, techniques combining both literature data and raw data will be discussed.

Learning Objectives

Those completing this course will have an understanding of the concepts and statistical methods required in pharmaceutical, biological, medical device and other health science research. They will be able to interpret results related to design and analysis issues of modern statistical techniques as routinely presented in the scientific literature and clinical trials. Course participants will be introduced to topic definitions followed by a discussion of their impact on clinical studies and the procedures for handling them in context, with many specific examples used.