Patient preferences for stroke prevention treatments in atrial fibrillation in Asia: A discrete choice experiment

Figures

• 

  Fig. 1. Example of choice task from the discrete choice experiment completed in 2023 by adults with atrial fibrillation in China, Taiwan, and South Korea (N = 307). Note: Participants were presented with 12 pairs of hypothetical scenarios and were asked to choose which one they would prefer.

• 

  Fig. 2. Preference estimates for non-vitamin K antagonist oral anticoagulant therapy in adults with atrial fibrillation in China, Taiwan, and South Korea (N = 307). Note: The mean marginal utility and 95 % CI of each attribute, relative to other attributes included in the DCE, are shown. Estimates are marginal utilities that denote the effect of deviating from a reference attribute level on preferences. Positive mean estimates with a 95 % CI >0 suggest participants place a higher value on the attribute level over the reference level, indicating that participants prefer the treatment option with this attribute level, all else being equal. *** indicates p < 0.001. Abbreviations: CI, confidence interval; MLE, maximum likelihood estimates; SE, standard error.

• 

  Fig. 3. Maximum acceptable risk of death over the next three years in adults with atrial fibrillation in China, Taiwan, and South Korea (N = 307). Note: The minimal improvement in death reduction required by patients to accept 1 % increase in poorer outcomes on other attributes is shown. MRS were estimated to determine the willingness of a patient to swap one attribute change for another providing equal utility. Abbreviations: CI, confidence interval; MLE, maximum likelihood estimates; MRS, marginal rates of substitution; SE, standard error.

• 

  Fig. 4. Predicted choice probability analysis of treatment preferences of adults with atrial fibrillation in China, Taiwan, and South Korea (N = 307) under different assumptions of fatal, disabling, and non-disabling event rates. Note: All lines denote the probability of respondents preferring treatment profile 1 (edoxaban-like; solid and dashed yellow lines) over profile 2 (rivaroxaban-like; solid and dashed blue lines) for varying levels of disabling events, non-disabling events, and fatal events (clinical performance is presented in Supplementary Table A.1). The shaded areas and the areas within the black dashed lines indicate 95 % confidence intervals. Each line summarises a change to baseline assumptions in either non-disabling events (solid orange and blue lines) and fatality rates (dashed blue and orange lines) for different ratios of mild and severe disabling events (shown in the three panels), holding all else equal.