Frequency and factors associated with placebo response in cancer-related fatigue treatment trials.

210 Background: There is limited literature on specific patient characteristics associated with response to placebo. The aim of this study was to identify frequency, and predictors of response to placebo. Methods: We conducted pooled analysis of the placebo arm in six randomized, double-blind, placebo-controlled clinical trials for cancer related fatigue. Baseline patient characteristics, symptoms (FACIT-F, and ESAS), Global Symptom Evaluation (GSE, rated as better, same and worse). Response was defined as an increase (D FACIT-F score) ≥ 3.5 points from baseline or change in ESAS of ≤ 1. Baseline patient characteristics and symptoms (as assessed by ESAS) were analyzed to determine their association with response to placebo using logistic regression analysis. ROC was calculated for GSE “better” of “4” or more (i.e., moderately better to very great deal better) and FACIT-F change. Results: Total of 306 patients with advanced cancer received placebo. The median age was 59, 54% were female, 76% were white, 51% were had college education or higher, the most common cancer type was genitourinary 24%. 55% had Zubrod performance ≤ 2. Median FACIF-F subscale score was 21. Placebo response was 176/306 (58%) with FACIT-F subscale ≥ 3.5 points and 185 (60%) with ESAS fatigue ≤ 1. Mean change in fatigue using FACIT- F subscale score was 7.18 (11.68) and ESAS was -2.21 (2.82). Change in FACIT-F and ESAS were significantly correlated (r = 0.56, P ≤ 0.001). 99/306 (32%) reported that their fatigue was “better” by GSE. Female gender (OR 2.72, P = 0.04), and anxiety (OR 1.39, P = 0.046) were significantly associated with placebo response. ROC with cut-off of GSE ≥ 4 and change in FACIT-F score was 11.5 and placebo response rate based on cut-off of 7.5 was 84/306 (27%). Conclusions: Our results confirm that placebo response is substantial in fatigue treatment trials in palliative patients. Placebo response rate varies based on the methods used to estimate response (from 27% to 60%). Female patients and higher anxiety at baseline were significantly associated with placebo response. More research is needed to accurately benchmark placebo response.