Ying Liu, Junlin Huang, Congli Chen, Jin Wu
In this pilot study, real-time remote monitoring emerged as the most consistent positive preference signal for continuous glucose monitoring device choice among parents of children with type 1 diabetes. Preferences for sensor replacement frequency appeared nonlinear, and estimates for other attributes were less stable. These findings are exploratory and hypothesis-generating, highlighting the need for larger multicenter studies with prespecified methods to examine preference heterogeneity and willingness to pay.
BACKGROUND: Continuous glucose monitoring is increasingly used in pediatric type 1 diabetes management, but families may value device characteristics differently and may trade off convenience, safety-related functions, remote data access, sensor replacement burden, and cost.
PURPOSE: This pilot discrete choice experiment explored parental preferences for continuous glucose monitoring device attributes among Chinese families of children with type 1 diabetes and examined the robustness of preference estimates under alternative respondent-inclusion and modeling strategies.
PATIENTS AND METHODS: Parents or legal guardians of children with type 1 diabetes completed an online discrete choice experiment. Each of ten core choice tasks presented two hypothetical continuous glucose monitoring devices and an opt-out option representing maintenance of the child's current monitoring strategy. Five attributes were included: measurement and calibration, low-glucose alert, remote monitoring, sensor replacement frequency, and monthly cost. The revised primary analysis excluded only responses with implausible or atypical completion times and retained respondents who failed the dominance task or consistently selected the opt-out option. Non-cost attributes were modeled categorically using effects-coded conditional logit models with cluster-robust standard errors at the respondent level. Sensitivity analyses applied alternative respondent-inclusion criteria.
RESULTS: Among 79 initial responses, 75 respondents were included in the revised primary analysis, contributing 750 core choice tasks and 2250 alternative-level observations. Real-time remote monitoring with automatic push notification showed a positive association with device choice in the revised primary analysis, although the estimate did not reach conventional statistical significance. This attribute remained positive and statistically significant in sensitivity analyses excluding dominance-task failures, excluding consistent opt-out respondents, and applying the original strict quality-control criteria. A 14-day sensor replacement frequency was positively associated with device choice in the revised primary analysis, whereas 30-day replacement did not show a consistent additional preference advantage. Measurement/calibration, low-glucose alert functions, and monthly cost showed less stable associations across analyses. Exploratory relative-importance analysis suggested that remote monitoring may be an important preference driver. Willingness-to-pay estimates were exploratory and should be interpreted cautiously because of imprecision in the cost coefficient.
CONCLUSION: In this pilot study, real-time remote monitoring emerged as the most consistent positive preference signal for continuous glucose monitoring device choice among parents of children with type 1 diabetes. Preferences for sensor replacement frequency appeared nonlinear, and estimates for other attributes were less stable. These findings are exploratory and hypothesis-generating, highlighting the need for larger multicenter studies with prespecified methods to examine preference heterogeneity and willingness to pay.