A comparison of two approaches for assessing patient importance weights to conduct an extended Q-TWiST analysis

Objective: Patient-centered methods for evaluating treatments require validated preference-elicitation techniques. We describe the validation of two preference-elicitation approaches for use in an Extended Q-TWiST treatment evaluation. The first method was an 'idiographic' approach, which attempts to capture intra-individual differences in the degree to which each domain distracted from and interfered with life activities. The second method, a Likert-scaled approach, asks patients to evaluate the importance of each quality-of-life (QOL) domain. Methods: Patient-reported QOL and preferences were assessed in participants with gastroesophageal reflux disease at baseline (n = 172), one week (n = 25), and 4 weeks after baseline (n = 100). Results: Both approaches demonstrated high internal consistency and the ability to discriminate known groups based on reported pain and number of days with symptoms. The idiographic approach exhibited responsiveness, although it was more highly correlated with QOL than the Likert-scaled approach. The Likert-scaled approach had good face validity but demonstrated low reliability compared to the idiographic approach. Conclusions: Both preference-elicitation methods exhibited promise as well as limitations. Future research should focus on increasing the reliability of the Likert-scaled approach, reducing the overlap between the idiographic approach and QOL, and examining the relationship between reliability and responsiveness for a range of illness trajectories.