PCR97 Discrete Choice Experiment Vs. Best-Worst Scaling: An Empirical Comparison in Measuring Treatment Preference of Lung Cancer Patients

To compare discrete choice experiment (DCE) and best-worst scaling (BWS) in eliciting lung cancer treatment preference measurement. A face-to-face survey was conducted among patients recruited using a quota sampling 2 tertiary hospitals China. Participants completed sociodemographic information, 13 DCE BWS tasks random order, the questions about difficulty of tasks. Each profile comprised six attributes, namely, overall survival (OS), risk severe adverse effects, severity pain, appetite, physical functioning, monthly cost. generalized multinomial logit model used to estimate attribute level weights, which were further compute relative importance (RI) for each attribute. The weights rescaled calculate correlation coefficient between two completion time, tasks, RIs compared BWS. total 161 survey, with 29.8%, 5.6%, 30.4% 34.2% at stage Ⅰ, Ⅱ, Ⅲ Ⅳ, respectively. majority male (69.6%) mean age 58.2 years. Respondents took longer complete than (25 min vs. 21min, P<0.001). However, task easier understand (74.5% 66.5%) answer (64.0% 57.8%), thus preferred by more respondents (49.7% 26.1%), generated similar rankings OS functioning ranked first 38% 24% DCE, 34% 17% BWS, Rescaled had strong 0.881 (P=0.000). ranking their correlated strongly. over