In the DCE, however, patients intuitively understood the task but had a hard time making a decision. For example, some patients wanted to place a cross on both sides of the AHP scale to indicate their strength of preference for both attributes in the comparison. The think-aloud statements revealed that patients initially had problems in understanding the AHP task (e.g., using the nine-point scale), but considered it easy to choose between two attributes and levels once understood. The question on perceived task difficulty resulted in 23% of patients reporting that the DCE was difficult or rather difficult to answer, whereas 18% found the AHP to be difficult or rather difficult. Moreover, the explorative subgroup analysis did not reveal differences in AHP weights or DCE coefficients depending on the order of questionnaire parts. The reported SDs for both AHP and DCE results suggest large heterogeneity in the collected preference data. In addition to this, the DCE results underline that patients wanted to avoid very severe but very rare systemic side effects more than other side effects, a result that was not as clear from the AHP. Improvement of visual function, two monthly monitoring schemes, approved drugs, less severe side effects, and injections on demand were (most) preferred relative to the respective reference attribute levels. The coefficients resulting from the dummy-coded regression analyses are to be interpreted in the following way: positive or negative estimates indicate a positive or negative added utility of the respective level relative to the utility associated with the reference level. The ranking of levels within each attribute was identical for AHP and DCE. In the AHP, patients rated the injection’s effect on visual function the highest, followed by the frequency of monitoring visits, approval status, side effects, and injection frequency. A detailed description of the AHP method is given in Saaty [ĪHP importance weights for attributes are presented in Table 2, together with local and global weights for levels in comparison with estimated coefficients for each attribute level from the DCE. Plausibility checks testing patients’ attention and task comprehension can nevertheless be performed in a DCE by including repeated choice sets or dominant options in the experiment. Such consistency measurement is not part of the DCE, which relies on the assumptions of random utility theory. The AHP permits calculation of a “consistency ratio” (CR), which measures how plausible one pairwise comparison is with respect to other pairwise comparisons. These weights do not per se convey information about the direction of a preference (positive or negative). Of note is the difference between the meaning of “preference” and “importance” in the two analytic approaches: coefficients in a DCE indicate a positive or negative preference for or against attribute levels in relation to the other levels of that attribute, whereas AHP importance weights are positive weight values for levels of an attribute and attributes, respectively, adding up to 1. AHP uses a direct mathematical approach to calculate importance weights for attributes, levels, and, if necessary, treatment alternatives, whereas DCE, being rooted in utility theory, estimates coefficients of a utility function using regression analysis. The expression of a preference for or against an option in DCE requires individuals to simultaneously weigh attribute levels against each other. Although the AHP asks respondents to value the relative importance of two attributes or levels directly on a nine-point ratio scale, the DCE requires a repeated discrete choice between options. DCE, however, uses combinations of attribute levels to develop descriptions of hypothetical treatment options and asks respondents to repeatedly choose between two or more of these options in order. AHP structures attributes and levels in a decision hierarchy, and respondents compare these to one another pairwisely at each level and in each cluster of the hierarchy. To facilitate the reading of this article, the terms “attributes” and “levels” are used throughout, despite often being referred to within the AHP methodology as decision “criteria” and “subcriteria,” respectively. 12/2016: Making Choices on the Journey to Universal Health Care Coverage: From Advocacy to AnalysisĭCE and AHP use very different approaches to measure the relative importance of treatment attributes and levels to decision makers.2/2017: “Mapping Studies” In Cost-Utility Analyses: New Recommendations From ISPOR Task Force.2/2017: ISPOR Releases New Task Force Recommendations for the Development of Clinician-Reported Outcome Assessments.
0 kommentar(er)
