Due to copyright issues, an electronic copy of the thesis must be ordered from the faculty. For the faculty to have time to process the order, the order must be received by the faculty at the latest 2 days before the public defence. Orders received later than 2 days before the defence will not be processed. After the public defence, please address any inquiries regarding the thesis to the candidate.
Trial Lecture – time and place
See Trial Lecture.
Adjudication committee
- First opponent: Professor Jack Bowden, University of Exeter, UK
- Second opponent: Assistant professor Rachel Knevel, Leiden University Medical Center, The Netherlands
- Third member and chair of the evaluation committee: Professor Hein Stigum, University of Oslo
Chair of the Defence
Professor emeritus Kristin Bjordal, University of Oslo
Principal Supervisor
Researcher Inge Christoffer Olsen, Oslo University Hospital
Summary
This thesis is situated in the field of biostatistics and causal inference, with applications in comparative effectiveness research in rheumatology. It addresses how to estimate causal treatment effects from observational data in the presence of unobserved confounding, particularly when treatments are categorical rather than binary.
The main research questions concern (i) which assumptions enable identification of causal effects with instrumental variables in multivalued treatment settings, (ii) how such effects can be interpreted and for which target populations they are valid, and (iii) how results can be generalized and extended to account for treatment effect heterogeneity.
Methodologically, the thesis develops a framework for instrumental variable analysis with categorical treatments, focusing on identification of local average treatment effects under plausible assumptions. It further proposes methods that combine instrumental variables with covariate adjustment to estimate conditional average treatment effects. The methods are evaluated through simulations and applied to observational data from the Norwegian NOR-DMARD registry, exploiting variation from the national drug tender system as an instrument.
Empirically, the thesis compares the effectiveness of tumor necrosis factor inhibitors as first-line biologic treatment for rheumatoid arthritis. The results indicate clinically meaningful differences in remission rates between specific treatments that are not detected using standard regression approaches. The work also demonstrates how accounting for effect heterogeneity can improve generalizability.
Overall, the thesis provides methodological advances and applied evidence showing that instrumental variable methods with categorical treatments can yield interpretable and policy-relevant insights when assumptions and target populations are carefully specified.
Additional information
Contact the research support staff.