Discrete choice experiment versus swing-weighting: A head-to-head comparison of diabetic patient preferences for glucose-monitoring devices

• Published in: PloS one Vol. 18; no. 7; p. e0283926
• Main Authors: Whichello, Chiara, Smith, Ian, Veldwijk, Jorien, de Wit, G Ardine, Rutten-van Molken, Maureen P M H, de Bekker-Grob, Esther W
• Format: Journal Article
• Language: English
• Published: United States Public Library of Science 28-07-2023
• Subjects: Biology and Life Sciences; Care and treatment; Case studies; Comparative analysis; Complications and side effects; Decision making; Devices; Dextrose; Diabetes; Diabetes mellitus; Diabetes mellitus (insulin dependent); Diabetes therapy; Diabetics; Engineering and Technology; Evaluation; Glucose; Glucose monitoring; Glucose monitors; Health aspects; Health education; Health literacy; Logit models; Measurement techniques; Medical equipment; Medical research; Medicine and Health Sciences; Monitoring; Patient compliance; Patient monitoring equipment; Patient outcomes; Patients; Physical Sciences; Preferences; Questionnaires; Rankings; Research and Analysis Methods; Risk assessment; Social Sciences; Surveys; Weighting; Netherlands; Taiwan
• ISSN: 1932-6203; 1932-6203
• DOI: 10.1371/journal.pone.0283926

Limited evidence exists for how patient preference elicitation methods compare directly. This study compares a discrete choice experiment (DCE) and swing-weighting (SW) by eliciting preferences for glucose-monitoring devices in a population of diabetes patients.
A sample of Dutch adults with type 1 or 2 diabetes (n = 459) completed an online survey assessing their preferences for glucose-monitoring devices, consisting of both a DCE and a SW exercise. Half the sample completed the DCE first; the other half completed the SW first. For the DCE, the relative importance of the attributes of the devices was determined using a mixed-logit model. For the SW, the relative importance of the attributes was based on ranks and points allocated to the 'swing' from the worst to the best level of the attribute. The preference outcomes and self-reported response burden were directly compared between the two methods.
Participants reported they perceived the DCE to be easier to understand and answer compared to the SW. Both methods revealed that cost and precision of the device were the most important attributes. However, the DCE had a 14.9-fold difference between the most and least important attribute, while the SW had a 1.4-fold difference. The weights derived from the SW were almost evenly distributed between all attributes.
The DCE was better received by participants, and generated larger weight differences between each attribute level, making it the more informative method in our case study. This method comparison provides further evidence of the degree of method suitability and trustworthiness.