MF9580 Epidemiological methods, beyond the basics

University of Oslo, Faculty of Medicine

Course content

The course introduces causal graphs (DAGs) and modern designs in epidemiology. It also introduces multilevel analysis and will give a user oriented introduction to handling of missing data (by multiple imputation) in Stata. It also covers the use of splines to handle non-linearity in regressions.

Learning outcome

The course has five main subjects:

• Causal graphs (DAGs)
• Modern Designs
• Multiple Imputation
• Splines in regressions
• Multilevel models

Also included are three minor subjects:

• Interaction
• Register epidemiology
• An introduction to Stata

Causal graphs introduce Directed Acyclic Graphs (DAGs) and will show examples of their use in medical research. DAGs are useful tools for understanding confounding, mediation and selection bias. A DAG analysis shows variables that should be adjusted for, and variables that should not be adjusted for.

The Design part will review classic research designs; cross sectional, cohort and case control. Focus will be on more advanced designs such as nested case control, case cohort and cross over designs. Strengths and weaknesses of the different designs will be discussed and examples of the different designs presented.

Multiple imputation: The lecture gives a short introduction to different missing-data mechanisms and how to handle missing data. The focus will be on multiple imputation, describing the concept of the method, how to choose a suitable imputation model, statistical inference, and challenges. Relevant Stata commands will be given and demonstrated through some examples.

Continuous variables and Splines: Categorizing continuous variables in a regression leads to loss of power. Instead, non-linear effects can be handled by using splines. We will provide Stata commands and examples of use.

Multilevel models gives an introduction to studies that lead to hierarchical data, what problem this gives in the analysis, and how to solve them. Focus is more on the interpretation of results than on the technical aspects of multi-level analysis.

Teaching

The course is organized as a full-time course over five days.

The three last sessions will include workshops using Stata. Student should bring own laptops. Students associated with UiO will get access to Stata (via UiO programkiosk). 

Examination

Home exam to be submitted after the course.

Admission to the course

Applicants admitted to a PhD programme at UiO sign up for classes and exam to this course in StudentWeb.

Applicants who are not admitted to a PhD programme at UiO must apply for a right to study before they can sign up for classes and exam to this course. See information here: How to apply for a right to study and admission to elective PhD courses in medicine and health sciences..

Back