Eliciting Patient-Centric Value Parameters: A National Best-Worst Scaling Profile Case Survey for Second-Line...

China faces the world's largest diabetes burden, with a 12.4% prevalence. A wide range of second-line therapies is available for patients with type 2 diabetes mellitus (T2DM), each with distinct characteristics. Although current guidelines advocate for a patient-centered approach in selecting second-line treatments, empirical evidence on Chinese patient's preferences for these therapies remains limited. We therefore elicited population‑level preferences using a Best-Worst Scaling profile‑case (BWS‑2) survey to inform decision making.

A seven-attribute instrument was designed through literature review, expert input, and a BWS object-case study used for attribute refinement (reported elsewhere). The final study employed BWS‑2 to elicit preferences across benefit, risk, and cost attributes. An orthogonal main‑effects design generated the choice sets. The survey was conducted through in-person interviews with patients with T2DM at healthcare institutions in three Chinese regions (eastern, central, and western). Data were analyzed using counting approach, conditional, and mixed logit regression. Preference heterogeneity was explored with subgroup analyses and latent class analysis (LCA).

Among 1517 respondents, counting approach and logit estimates were consistent; most attribute coefficients were statistically significant in logit analysis, with patients showing the highest preference for reduction in HbA1c, cardiovascular protection, and zero out-of-pocket costs compared with the reference levels. Weight change and mode of administration exerted modest influence. LCA identified six unobserved subgroups. Preferences heterogeneity were also explored by splitting the sample according to patient urban-rural residence and income characteristics.

Chinese patients with T2DM prioritized HbA1c reduction, cardiovascular benefits, and cost, while weight change was rated as the least important attribute. LCA further revealed preference heterogeneity, including limited concern for adverse events and neutral attitudes toward all attributes.