Description: In this course, faculty will present and discuss considerations related to threats to the validity of research on pregnancy pharmacoepidemiology. The course will consist of three recorded presentations by faculty, each of them 30’-40’ long, and a live 30’-60’ Q&A session. This is a new format for this course, in line with all skill courses offered as a part of ICPE All Access 2020. In 2015, the course provided an overview of issues that are particular to drug safety research in pregnancy. In 2016 the course focused on challenges related to exposure measurement, in 2017 the focus was on maternal and neonatal outcomes, in 2018 the course presented pros and cons of various study designs. In 2019, the course discussed significance testing and analytical approaches to address biases. The content of the 2020 edition of this course will cover some of the main themes discussed in previous years. The course will start with a general overview on drug safety in pregnancy, planned to provide context to the course and to be especially useful for first-time course attendees; followed by more focused talks on unmeasured confounding, selection bias, random error and external validity. The live Q&A session will take place on October 14, 2020, starting at 9 am EST.
To be better equipped to critically assess the validity of studies on drug safety in pregnancy To understand key aspects of study design related to exposure, outcome, confounding, generalizability and random error To be able to identify useful approaches to address potential threats to the validity of studies on drug safety in pregnancy, and to recognise the circumstances when they are most appropriate. Outline: Threats to the validity of studies on drug safety in pregnancy. In the first lecture, an overview will be provided of common threats to the validity of studies on drug safety in pregnancy. Design and analytic approaches that can help to address potential misclassification bias, confounding, and selection bias will be introduced. The aim of this first lecture is to provide students a general appreciation of the scope of approaches available before delving into specific issues in the subsequent lectures. Instructor: Brian Bateman, Brigham and Women’s Hospital, Harvard Medical School. Approaches to mitigate unmeasured confounding. In this session, we will discuss approaches to mitigate unmeasured confounding in pharmacoepidemiology pregnancy safety research. The focus will be on those aspects of unmeasured confounding that are specific to pregnancy research (e.g., variables or characteristics that typically remain unmeasured in pregnancy-related research, unique analytic approaches available to address unmeasured confounding). Planned topics for discussion include propensity score calibration, adjustment with data from externals sources, and sibling analysis. The approaches will be illustrated using pregnancy-related examples. Instructor: Andrea Margulis, RTI Health Solutions Challenges related to selection bias, random error, and external validity. In this session, we will address other potential threats to the validity of perinatal pharmacoepidemiologic studies. These include (1) challenges arising when restricting the study cohort to live births; (2) the degradation of quantitative information when focusing on statistical significance testing; and (3) common concerns around external validity. The focus of the lecture will be on approaches that can be used to address these concerns. Instructor: Krista Huybrechts, Brigham and Women’s Hospital, Harvard Medical School