Patient selection in target trial emulation using data from Cancer registry of Norway

Abstract

Background: Breast cancer is a significant public health challenge with diverse treatment outcomes. This underscores the need for better patient selection in clinical trials to improve representativeness and treatment efficacy. This thesis uses data from the Cancer Registry of Norway (CRN) to assess how modifications to eligibility criteria, as outlined in the PALOMA-II randomized controlled trial, can influence the number of patients enrolled, demographic representation, and survival outcomes. Method: This study adopts a retrospective cohort study design, analyzing CRN data from 2015 to 2021. By utilizing combinatorial filtering techniques, the study creates three distinct patient cohorts—broad, neutral, and strict—each defined by varying degrees of selection stringency. The method involves extensive data manipulation and survival analysis, with Kaplan-Meier curves generated to compare survival probabilities across these cohorts. Additionally, demographic characteristics and clinical profiles are examined to understand the implications of each selection strategy. Results: The findings reveal that broader eligibility criteria result in patient cohorts whose survival probabilities and demographic distributions closely align with the general breast cancer population recorded in the CRN. This suggests that less restrictive selection criteria enhance the generalizability of trial results and may improve patient recruitment and representativeness. In contrast, strict selection criteria, while potentially increasing the homogeneity of trial participants and the specificity of outcomes, significantly limit the diversity of the study population and may reduce the external validity of the research findings. Conclusion: Modifying patient selection criteria in clinical trials has impacts on study representativeness and the applicability of research findings to clinical practice. It advocates for a balanced approach in clinical trial design, suggesting that eligibility criteria should neither be too restrictive, to avoid excluding significant patient subgroups, nor too broad, to prevent dilution of potential treatment effects. Future research should explore adaptive trial designs and further validate these findings across other diseases and populations to enhance the translational impact of clinical research. This study underscores the critical role of patient selection in clinical trial design and its consequential impact on health outcomes in breast cancer treatment.