Preference Phenotypes in Thyroid Nodule Management: A Patient Segmentation Approach

Objective: Patient preferences regarding thyroid nodules are poorly understood. Our objective is to (1) employ a discrete choice experiment (DCE) to explore risk tradeoffs in thyroid nodule management, and (2) segment respondents into preference phenotypes. Study Design: DCE. Setting: Thyroid surgery clinic, online survey. Methods: A DCE including 5 attributes (cancer risk, voice concerns, incision/scar, medication requirement, follow-up frequency) was refined with qualitative patient and physician input. A final DCE including 8 choice tasks, demographics, history, and risk tolerance was administered to participants with and without thyroid disease. Analysis was performed with multinomial logit modeling and latent class analysis (LCA) for preference phenotyping. Results: A total of 1026 respondents were included; 480 had thyroid disease. Risk aversion was associated with increasing age (P <.001), female gender (P <.001), and limited education (P =.038), but not previous thyroid disease. Cancer risk most significantly impacted decision-making. Of the total possible utility change from thyroid nodule decision-making, 47.8% was attributable to variations in cancer risk; 20.0% from medication management; 14.9% from voice changes; 12.7% from incision/scar; and 4.6% from follow-up concerns. LCA demonstrated 3 classes with distinct preference phenotypes: the largest group (64.2%) made decisions primarily based on cancer risk; another group (18.2%) chose based on aversion to medication; the smallest group (17.7%) factored in medication and cancer risk evenly. Conclusion: Cancer risk and the need to take medication after thyroid surgery factor into patient decision-making most heavily when treating thyroid nodules. Distinct preference phenotypes were demonstrated, reinforcing the need for individual preference assessment before the treatment of thyroid disorders.