Assessing an ASCO Decision Aid for Improving the Accuracy and Attribution of Serious Adverse Event Reporting From Investigators to Sponsors.

PURPOSE Investigators often send reports to sponsors that incorrectly categorize adverse event (AE)s as serious or attribute AEs to investigational drugs. Such errors can contribute to high volumes of uninformative investigational new drug safety reports that sponsors submit to the US Food and Drug Administration and participating investigators, which strain resources and impede the detection of valid safety signals. To improve the quality of serious AE (SAE) reporting by physician-investigators and research staff, ASCO developed and tested a Decision Aid. METHODS A preliminary study with crossover design was conducted in a convenience sample. Physician-investigators and research staff were randomly assigned to receive case studies. Case studies were assessed for seriousness and attribution, first unassisted and then with the Decision Aid. Participants completed a feedback survey about the Decision Aid. Effectiveness of reporting and attribution are reported as odds ratios (ORs) with 95% CI. Power to detect associations was limited because of a small sample size. RESULTS The Decision Aid did not significantly affect accuracy of determining seriousness (OR, 0.87; 95% CI, 0.31 to 2.46), but it did significantly increase accuracy of attributing an SAE to a drug (OR, 3.60; 95% CI, 1.15 to 11.4). Most of the 29 participants reported that the Decision Aid was helpful (93%) and improved decision-making time (69%) and confidence in reporting (83%), and that they would use the Decision Aid in practice (83%). CONCLUSION The Decision Aid shows promise as a method to improve the quality of SAE attribution, which may improve the detection of valid safety signals and reduce the administrative burden of uninformative investigational new drug safety reports. Study of the Decision Aid in a larger sample with analysis stratified by participant role and SAE reporting experience would further assess the tool's impact.