Advanced Pharmacoepidemiology Methods: Mitigating Missing Data Concerns with External or Internal Subset Validation Populations

Thursday, August 25, 2022

2:00 PM – 6:00 PM CEST

Location: C1M1

Publication Number: 19

Course Aim This course will introduce participants to the methods available to mitigate concerns around missing data in Real World Evidence studies. These include the identification and incorporation of data from external validation groups, the use of subset internal calibration groups, and the use of multiple imputation methods when data are believed to be missing at random. Real-world examples will be used across the lecture sessions to illustrate and connect different concepts.

Requisites Statement: This is an entry level course.

Course Objectives

1. Describe the potential analytic techniques or other solutions when studying populations for which data are missing for entire comparator groups or when key variables are missing for portions of both treatment and controls groups in an overall study population.

2. Appreciate the present state of use of Real World Data / secondary use observational data as an adjunct to primary use data collected in the course of a clinical trial or registry study. Understand the appropriate incorporation of these data as external controls using rigorous techniques to control for biases.

3. Understand analytic methods to account for partially missing data (e.g., when BMI is missing for 40% of the study population) in an overall study.

4. Appreciate some of the recent trends in overcoming missingness including the use of synthetic data.

Syllabus Outline The course will be live followed by a panel Q&A session.

1.Landscape Analysis on Techniques to Deal with Missing Data (Janick Weberpals): This session will discuss the landscape of analytic techniques to deal with missing data including statistical techniques like multiple imputation, and new approaches that involve sourcing external validation groups or using synthetic data.

PRE-READ: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8168830/

2. External Control Arms to Support Drug Applications/Safety Analyses (Lourens Bloem): For single-arm trials (e.g., cancer or rare diseases) or uncontrolled extensions of late-stage randomized controlled trials, there is often a need to add observational data to compensate for a “missing” comparator group using a variety of methods. As these “real world data” are added, careful attention must be paid to appropriate design and analytical approaches to mitigate bias. This session will focus on techniques to maintain scientific rigor in these approaches.

3. External Validation Groups to Support Pregnancy Registries (Corinne De Vries): In order to monitor the safety of drugs in pregnant women, manufacturers are often required to create pregnancy registries. Incorporating appropriate comparator groups in these registries is challenging. It is often necessary to find external validation groups to contextualize the data in the registries. This session will focus on techniques to incorporate these external validation groups into pregnancy registry studies.

4. Subset Calibration Analyses (Jenny Sun): Pharmacoepidemiologic studies are increasingly conducted within linked databases, often to obtain richer confounder data. However, these data are often missing, but potentially only on a portion of the population. Yet, when these data are only available on a subset of patients, there is a strong potential for selection bias. This session will focus on techniques to adjust for potential selection biases when using subset analyses using inverse probability of selection weights (IPSW) in conjunction with inverse probability of treatment weights (IPTW).

PRE-READ: https://pubmed.ncbi.nlm.nih.gov/35015823/

Lead Speaker(s)

JM

Judith C. Maro, PhD

Harvard Pilgrim Health Care Institute
Boston, Massachusetts, United States

Slides

Speaker(s)

Lourens T. Bloem, PharmD, PhD

Assistant Professor Clinical Therapeutics
Division of Pharmacoepidemiology and Clinical Pharmacology, Utrecht Institute for Pharmaceutical Sciences, Utrecht University
Utrecht, Utrecht, Netherlands

Slides
Jenny Sun

Pfizer, Inc. Worldwide Medical and Safety
New York, New York, United States

Slides
Cd

Corinne de Vries

Head of Translational Sciences
EMA
Amsterdam, Netherlands

Slides
Janick Georg Weberpals, RPh, PhD

Instructor in Medicine
Brigham and Women's Hospital, Harvard Medical School
Boston, Massachusetts, United States

Slides