PhD defence Camila Olarte Parra

12-10-2020 from 15:00 to 17:00
Aula Ceremoniezaal, Campus Aula, Voldersstraat 9, 9000 Gent
Camila Olarte Parra
Add to my calendar

PhD defence Camila Olarte Parra

Camila Olarte Parra is defending her PhD in statistical data analysis entitled 'Causal inference from observational studies in clinical research: lessons learned from kidney transplantation and beyond'.

Promotors are prof. dr. Els Goetghebeur, Ghent University and prof. dr. Aeilko Zwinderman, University of Amsterdam, the Netherlands.

The defence can be followed online: 0597c97f?lti-scope=d2l-resource-syncmeeting-list

Summary of the thesis:

To inform clinical decisions, we need evidence on causal effects of possible interventions. The gold standard for causal effect estimation is a randomised controlled trial (RCT). When RCTs are not feasible, ethical, or timely, we must rely on observational studies. Typically, the treated and control groups will differ in prognostic characteristics. Ignoring these differences, can lead to considering that an observed effect is due to the treatment when it is rather determined by these other characteristics. In this doctoral work, we chose kidney transplantation as a casestudy to illustrate the potential pitfalls and biases that arise when deriving causal effects from observational studies. We first performed a systematic review to identify all studies that compared pre-emptive kidney transplantation (i.e. without dialysis) vs dialysis possibly followed by transplantation for end-stage kidney disease. We identified major limitations of previous studies, particularly in selection of participants and setting time-zero i.e., start of follow-up. Using the nationwide Swedish Renal Registry, we embarked on the quest of answering the causal question while illustrating how to avoid the sources of bias identified. We then focused on how causal claims emerge in study reports of observational studies in general. When reporting the findings, the proper use of causal methods and the corresponding causal language is critical to avoid misleading conclusions.