Use of secondary subgroup analysis of clinical trials across solid malignancies.

Abstract

e13618 Background: The use of secondary (subgroup) analyses in medical literature is increasing. Findings from subgroup analyses may be misleading given they are observational in nature and do not resemble a randomized comparisons. Our study aimed to characterize the use of post hoc subgroup analysis in interventional clinical trials across solid malignancies. Methods: We searched OVID database for articles subgroup analyses from the year 2000 until Feb 15, 2022. Some of the used search terms were post-hoc, subgroup, secondary, cancer, tumor, breast, colorectal. 11641 articles were screened to include all secondary analyses of interventional solid tumor clinical trials. We excluded original trials, secondary analyses of cohort/observational studies. Results: Our final analysis included 1491 articles. Most studied malignancy was gastrointestinal cancers (22.6%), followed by genitourinary cancers (21%), and breast cancer (19%). The number of papers published per year increased steadily over the years with a peak in 2021 (16% of all articles). About 78% of articles reported receiving funding, with 40% reporting pharma-related funding. 29% of articles reported the usage of a medical writer. 55% of articles reported that authors had conflicts of interest (COI). Pharma-related funding was significantly associated with usage of a medical writer (p < 0.001). The mean percentage of authors with COI was 31.8 ± 0.95%. Articles that reported usage of a medical writer had a higher percentage of authors with COI (p < 0.0001). Women represented 32.7% and 25.9% of first and last authors, respectively. About 44% of both first and last authors resided in North America. Half of the articles explicitly indicated that they were secondary analyses in their titles. [Table]. Conclusions: Our study indicated that the use of secondary subgroup analysis in solid malignancies clinical trials is increasing. Half of the studies did not clarify the nature of them being secondary analyses in the title. Given their increased use, caution must be entertained when interpreting the results of secondary analysis. [Table: see text]