Analysis of adverse events attribution and reporting in cancer clinical trials: A systematic review.

Abstract We systematically reviewed the quality of AEs reports in published oncology trials analyzing also the bias in the attribution process. We searched MEDLINE, PubMed (2000-2019) selecting randomized, double-blind, placebo-controlled, and phase 3 cancer trials using exclusively targeted therapy or immunotherapy-related drugs. The proportion of publications with complete AE reports (including both all-cause and drug-related AE data) and the AEs attribution ratio (patients with drug-related over all-cause AE) were investigated. Among 60 trials (38,174 patients) included, 40 (66.6%) presented an incomplete report of AEs attribution. Journals with the lowest impact factor were significantly associated with deficient reports of grade 3-4 AEs (p = 0.02). Under placebo administration, the median incidence of all-grade drug-related AEs was 49% (IQR 39-56). The median attribution ratio for all-grade AEs in the active and placebo arms was 88.9% (IQR 79.8-93) and 53.9% (IQR 43.4-60.9), respectively. The AEs reporting and attribution process appear to be more unreliable than expected.