Subgroup analyses are essential to generate new hypotheses or to estimate treatment effects in clinically meaningful subgroups of patients. They play an important role in taking the next step towards personalized surgical treatment for brain tumor patients. However, subgroup analyses must be used with consideration and care because they have significant potential risks. Although some recommendations are available on the pearls and pitfalls of these analyses, a comprehensive guide is lacking, especially one focused on surgical neuro-oncology patients. This paper, therefore, reviews and summarizes for the first time comprehensively the practical and statistical considerations that are critical to this field. First, we evaluate the considerations when choosing a study design for surgical neuro-oncology studies and examine those unique to this field. Second, we give an overview of the relevant aspects to interpret subgroup analyses adequately. Third, we discuss the practical and statistical elements necessary to appropriately design and use subgroup analyses. The paper aims to provide an in-depth and complete guide to better understand risk modeling and assist the reader with practical examples of designing, using, and interpreting subgroup analyses.
Keywords: Confounding; Malignant glioma; Multiplicity; Study design; Subgroup analysis.
© The Author(s) 2024. Published by Oxford University Press on behalf of the Society for Neuro-Oncology.