Meeting: 2016 ASCO Annual Meeting
Title: Adverse event (AE) attribution in symptom intervention and cancer prevention trials.



Background: Assignment of AE attribution (as unrelated, unlikely, possibly, probably, and definitely related to the treatment) adds time and cost to the trial process. External factors have been shown to affect attribution reporting. Hillman et al. (2010) showed that 50% of AEs in the placebo arms of two multicenter cancer treatment trials were reported as related to the study drug. Here, we evaluate the patterns of AE attribution in symptom intervention (cancer patients) and cancer prevention (participants at high risk for cancer) setting. Methods: Seven multicenter placebo-controlled trials were analyzed separately (1253 patients, 7064 AEs). Frequency and severity of AEs were summarized by arm. Attribution and percentage of repeated AEs whose attribution changed overtime were summarized for the placebo arms. Percentage of physician over- or under-reporting of AE relatedness was calculated for the treatment arms using the placebo arm as the reference. Results: Thefrequency and severity of AEs observed in treatment and placebo arms were similar. 12 57% of AEs in the placebo arm were reported as at least possibly related to the treatment. Attribution changed overtime in 0 34% of repeated AEs. Commonly over-reported AEs include fatigue, constipation, diarrhea, and GI disorder. Conclusions: Up to50% of AEs in the placebo arms were classified as possibly related to the treatment. This combined with up to 34% change in attribution of repeated AEs in the same patient suggests that assignment of AE attribution in this setting is also unreliable and subjective. Symptom interventionPreventionDiseaseVariousBarretts esophagusBronchial dysplasiaColorectalInterventionEpoetin AlfaIronAspirinMetforminSulindacMyoinositolAtorvastatin, Sulindac, ORAFTI Synergy1# patients (placebo)333 (165)490 (163)123 (30)77 (36)60 (30)85 (41)85 (22)# AEs in placebo, treatment1378, 13791085, 204431, 4646, 71145, 121211, 26384, 160% grade 3 5 in placebo, treatment14, 1633, 260, 41, 03, 31, 35, 4% relatedness in placebo (possibly, probably + definitely)12 (10, 2)24 (21, 3)55 (48, 7)57 (50, 7)19 (19, 0)29 (29, 0)37 (35, 2)% repeated AEs with attribution change in placebo25 (130/527)34 (149/439)0 (0/15)0 (0/5)20 (9/46)11 (8/76)8 (2/26)