Applied Sensitivity Analyses

It is widely recognized that sensitivity analyses of study design choices and analytic assumptions ‎help to interpret the robustness of pharmacoepidemiology studies. To encourage increased use ‎of well-described techniques, this workshop will provide an introduction and demonstrations of a ‎range of sensitivity analyses typically applied in pharmacoepidemiology using healthcare ‎database studies.‎

Part 1: Provides a brief introduction to principled pharmacoepidemiology with healthcare ‎databases and introduces the logic of sensitivity analyses during the study design phase and ‎post-hoc quantitative bias analyses of effect estimates. We explain the value of study design ‎diagrams and directed acyclic graphs in this process. We conclude by introducing a case study to ‎be used in Parts 2 and 3.‎

Part 2: Uses a case study to work through several sensitivity analyses of study design choices ‎by introducing typical variations in study designs, including choice of comparator, defining ‎discontinuation of medication use (i.e., allowable gap), and variations in exposure risk window ‎length, covariate assessment window length, duration of minimum induction period, and follow-up ‎model (as-treated analysis with censoring that accounts for treatment changes vs. ITT analysis). ‎All sensitivity analyses will be conducted with audience input based on the case study introduced ‎in Part 1.‎

Part 3: Focusing on confounding we use DAGs to explain the concept of negative/positive control ‎outcomes and discuss examples. This is followed by quantitative confounding bias analysis ‎testing the influence of external assumptions or outside data on our understanding of residual ‎confounding in the specific study setting. All sensitivity analyses will be conducted with audience ‎input based on the case study introduced in Part 1.‎

This 3.5-hour pre-conference course focuses on principles and concepts, not on mathematical ‎details. The course faculty includes members from academia, industry and regulatory to teach a ‎broad perspective on the uses and interpretation of sensitivity analyses.‎

Educational objectives
Upon completing this course, participants will be able to:‎

a. Recognize the value and applicability of sensitivity analysis in pharmacoepidemiology research
b. To discuss the findings of sensitivity analyses of study design choices and understand the ‎impact of varying study design assumptions in practical examples
c. To explain quantitative confounding bias analysis and understand their interpretation in specific ‎examples
d. To develop strategies on how to plan for and communicate results of sensitivity analyses from ‎the perspective of various stakeholders

Target audience

This course is designed for scientists, regulators, coverage decision makers conducting and ‎interpreting pharmacoepidemiology research studies using healthcare databases

This beginner-to-intermediate-level course assumes knowledge of basic pharmacoepidemiologic ‎methods.