Elsevier, Journal of Cancer Policy, Volume 30, December 2021
Introduction: Women continue to be underrepresented in oncology clinical trials, leading to poor, underpowered subgroup analyses that cannot be generalized to cancer patients in practice. In 2014, the US Food and Drug Administration (FDA) released an Action Plan, which included actions to improve the quality and reporting of demographic subgroup data. We sought to evaluate the five-year progress since the release of this report by assessing the credibility of sex-specific subgroup analyses in oncology clinical trials. Methods: We reviewed the FDA Hematology/Oncology Approvals website for New Molecular Entities (NMEs) that were approved for adults from 2015 to 2020. Publications and their supplementary indexes were reviewed by two authors (K.J. & A.R.) against ten criteria that gauge the credibility of subgroup analyses by assessing factors related to study design, analysis, and context. One point was awarded for each criteria met, for a maximum score of 10. Results: We identified a total of 73 NMEs approved for cancer treatment between 2015–2020, of which 61 met our eligibility criteria. Of these, 32 studies (52 %) reported a subgroup analysis by sex and were included in our analysis. Phase 2 (41 %) and Phase 3 (53 %) studies represented most studies. No study met ≥3 credibility criteria. Conclusion: Only half the studies included in our analysis reported outcomes by sex, which suggests the activities stipulated in the 2014 US FDA Action Plan might be ineffective. This is concerning as uncredible sex-specific subgroup analyses can lead to wrongful clinical decision-making and poor patient outcomes. Policy summary: Our findings suggest sex-specific subgroup analyses in oncology are not credible and users of these data should interpret results with caution. Regulatory bodies, such as the US FDA, ought to mandate subgroup analyses by demographic groups in drug applications. Peer-reviewed journals could ensure investigators disclose study results by sex as a condition for publication.