University of Leeds
Trial design and analysis methods to incorporate the complexities of surgery
Abstract
dc:description.abstractBackground: Current surgical trial design targets the average treatment effect (ATE) to compare two procedures while averaging over surgeon experience. This approach is potentially misleading. The importance of estimating the ATE conditional on surgeon experience (CATE) is recognised, but current guidance is limited to analysis methods, and current design approaches do not ensure the robustness of those analyses. Primary Research Question: What can be done in the design of a surgical trial to facilitate the estimation of the CATE? Methods: Drawing from Design of Experiments methods, I derive an orthogonal design targeting the CATE where surgeon and experience are non-randomisable. I apply this design to practical scenarios and explore how the precision of the CATE estimator could be maximised. I also compare properties of the CATE estimator under the orthogonal design to the current standard design. Findings: The orthogonal design necessitates conditions on surgeon and experience, and randomises patients to procedure stratified by their surgeon-by-experience class. Sample size, number of surgeons and spread of experience all affect the precision of the CATE. Strategies for maximising precision of the ATE employed in current designs are insufficient for ensuring precision of the CATE. The optimal orthogonal design substantially outperforms the current standard design in terms of precision of the CATE in the scenarios explored. Strategies ensuring sufficient precision of the CATE include manipulation of the spread of experience via purposeful recruitment of surgeons, including managing the number of surgeons to include and which experience levels to prioritise. Conclusion: There is huge scope for improvement upon current approaches to surgical trial designs regarding the precision of the CATE estimator. The orthogonal design derived here facilitates estimation of the CATE, and should be emulated as closely as possible in practice. Furthermore, there are numerous recruitment strategies that could improve precision of the CATE.
Degree
thesis:*- Name dc:type.qualificationname
- Ph.D
- Level dc:type.qualificationlevel
- doctoral
- Grantor dc:publisher.institution
- University of Leeds
- Year dc:date.issued
- 2023
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Corrigan, Neil Paul
- Advisors dc:contributor.advisor
-
- Walwyn, Rebecca
- Brown, Julia
- Emsley, Richard
- Jayne, David
- ORCID person:identifier.orcid
Rights
- Language dc:language.dcterms_iso639_2
- eng
Identifiers
dc:identifier.*- Dc Identifier Grantnumber
- DRF-2018-11-ST2-079