Skip to content

Epidemiological Methods for Causal Mediation Analyses

Course Number
5400 (Epidemiology)
Course Instructor(s)
Peter M Smith, Brice Batomen Kuimi

Course Description

Examining the chain of relationships between measures is central to many observational investigations. For example, researchers may be interested in the relative importance of different pathways through which an independent variable might impact on an outcome, in order to determine which variables have the greatest potential for future interventions. A variable located on a causal pathway between two variables is referred to as a mediating variable. While methods to examine mediation have been around for decades in the social and psychological sciences they have not been extensively used in the field of epidemiology until the last decade. In addition, common methods to assess mediation (e.g. using a series of regression models with and without the mediating variable of interest) are not always appropriate when the outcome of interest is not normally distributed (e.g. for dichotomous and time-to-event outcomes), or when exposure-mediator interaction is present.

Course Objectives

The purpose of this course is to introduce students to conceptual issues and analytical approaches to assess causal mediation effects under observational study designs. At the completion of the course students should have  developed an understanding of formulating research questions in causal terms, the merits of different statistical approaches to assess mediation, and the theoretical and statistical assumptions behind these approaches. In turn, they will be able to determine the most appropriate approach to assessing mediation, given their data source, available measures and research question.

The course will begin with an overview/review of key concepts in causation and mediation. We will then introduce methods to derive directed acyclic graphs (DAGs) for research questions, and briefly outline how DAGs can be used to better understand questions of bias (including confounding) and mediation. The course with then focus on approaches to assess mediation with a main focus on path and structural equation modelling, and potential outcomes (counterfactual) approaches. Path models/SEM and causal models specified in terms of potential  outcomes allow researchers to quantify and estimate the total effect, and the indirect and direct effects, between an independent variable and an outcome of interest. However, they differ in important ways, which will be discussed as the course progresses.

Methods of Assessment

Homework Problem Sets 45%
First Assignment (a one-page summary that outlines their specific mediation question of interest, with an accompanying DAG). 25%
Final assignment. Students will extend their first assignment (DAG model and one page summary) to outline and justify a proposed analytical approach to examine their research question. 30%

General Requirements

This course is intended for graduate students with an interest in intermediate and advanced quantitative research methods and data analyses. Students taking the course will require an intermediate understanding of confounding, causality and the principles of measurement such as reliability and validity, as this knowledge is assumed in the course materials. Students will also require at least intermediate skills in statistical programing, in order to complete course homework assignments, and understand examples given throughout the course lectures.
Specific pre-requisites are:

▪ The completion of one advanced course in epidemiological study designs (e.g. CHL5406H – Quantitative Methods for Biomedical Research) or an equivalent course in observational study
design taught in other disciplines such as health services research, biostatistics or health economics (e.g. HAD 6770H – Applied Research Methods; HAD 5763H – Advanced Methods in
Health Services Research).
▪ At least one intermediate or advanced course in statistics or biostatistics including multivariable regression models for continuous and non-continuous (e.g. binary) outcomes (e.g.
CHL5202H – Biostatistics II)