Patient Preferences for Diabetes Treatment Attributes and Drug Classes.

Objective: To identify which treatment attributes are most influential in determining patient preferences for diabetes treatments and explore patient preferences for diabetes drug classes. Research design and methods: US adults with type 1 or type 2 diabetes completed an online adaptive conjoint analysis survey. The survey examined 14 attributes, including efficacy, regimen, and risk of common side effects and rare but serious adverse events. Respondents selected between hypothetical treatments with different attributes. Sawtooth Software, ordinary least-squares regression, and hierarchical Bayes regression were used to calculate utilities (i. e. preference weights), importance ratings, and shares of preference across 13 diabetes drug classes or combination products. Results: A total of 167 adults (mean age 58 years; 55% female) with type 1 or type 2 diabetes completed the survey. Based on importance ratings, the most influential attributes driving preferences were regimen, risk of diarrhea, weight change, risk of hypoglycemia, and efficacy. Sodium-glucose co-transporter-2 inhibitors (SGLT-2is) were highly preferred in direct comparison to each of the other classes (range: 84.2-99.9%), with the exception of dipeptidyl peptidase-4 inhibitors (DPP-4is); DPP-4is (52.9%) were preferred over SGLT-2is (47.1%). Conclusions: Although preferences varied across participants, attributes with the greatest likelihood of affecting daily life and routine were generally more influential in determining patient preferences. DPP-4is and SGLT-2is were overwhelmingly preferred over other drug classes, primarily due to favorable regimen and side effect profiles. Understanding patient preferences can help optimize patientcentered treatment and may lead to improved patient satisfaction, adherence, and outcomes. Limitations: The primary limitations of this study are that a small sample size of type 1 diabetes patients were included, which may reduce the reliability of the preference estimates, and patients were recruited from a patient panel and may not be representative of patients with diabetes in the US.