How Well Can Analytic Hierarchy Process be Used to Elicit Individual Preferences? Insights from a Survey in Patients Suffering from Age-Related Macular Degeneration

Marion Danner*, Vera Vennedey, Mickael Hiligsmann, Sascha Fauser, Christian Gross, Stephanie Stock

*Corresponding author for this work

Research output: Contribution to journalArticleAcademicpeer-review


Background In this study, we tested the feasibility of an interviewer-assisted analytic hierarchy process (AHP) in a special patient population with age-related macular degeneration (AMD). Objectives One aim was to generate preference weights regarding AMD treatment characteristics. A secondary aim was to explore the consistency of preference judgments and reasons for inconsistency. Methods We generated quantitative importance weights for decision criteria using the matrix multiplication method. A qualitative study component in the form of asking patients to think aloud throughout their judgments was implemented to facilitate understanding of quantitative findings. Consistency ratios were calculated as a measure of logical judgment performance within AHP. If consistency ratios exceeded 0.2, we explored reasons for inconsistency. Results We interviewed 86 patients and generated preference weights for criteria. Patients rated the injection's effect on visual function the highest (0.44), followed by the frequency of monitoring visits (0.18), approval status (0.13), injection frequency (0.13), and side effects (0.12). Inconsistency in judgments was prevalent at the subcriteria level. Whereas much of the observed inconsistency was due to an excessive use of high/extreme value judgments, these judgments seemed to result from patients reasonably trying to highlight their strong preferences. Conclusion Our study combines quantitative with qualitative data to explore patients' preference weights and decision processes using the AHP. It suggests that the type of inconsistency observed in judgments of AMD patients mostly results from rational decision making, not from error or lack of understanding. Further research should address which type and extent of inconsistency might be acceptable in different AHP settings.
Original languageEnglish
Pages (from-to)481-492
JournalThe Patient: patient-centered outcomes research
Issue number5
Publication statusPublished - Oct 2016

Cite this