Skip to Content

Understanding treatment mechanisms in clinical survival trials - a causal approach

Research Objectives:

Survival analysis is a very important method in statistics, especially in medicine, with thousands of published applications every year. The standard approaches focus on analysing the risk of an event occurring, e.g. the risk death if the aim of the study is to prolong the life of patients with a serious disease. What is typically missing in standard clinical survival studies is to use efficiently all the clinical observations and measurements that are made of the patients during the trial. These are almost never integrated into the survival analysis. There are new methods that can do this, e.g. dynamic path analysis, and this could teach us a lot about how and why the treatments work, and about which patients really are helped by the treatments. So the point is to find out how to learn more from the survival data collected in large clinical trials and epidemiological studies. This is also connected to new field of causal inference and the estimation of mediation and direct and indirect effects which is recently a very popular subject.

Description of work:

The project will start with analysing data from a large clinical trial which concerned patients who have had a myocardial infarction and where one wished to treat these so as prevent a new infarction or death. This is a study based in Norway with almost 9000 patients. Next some methodological work will be done. This will take up challenges that were met in the first analysis, and also contribute to the further developments of methods like dynamic path analysis and causal mediation analysis. There are lots of interesting methodological issues that can be taken up, dependent on the interest of the candidate.
In addition to researchers in Oslo, Theis Lange from the University of Copenhagen will be co-supervisor on the project.

Host Institution: University of Oslo

about seo