Measuring patients' priorities using the Analytic Hierarchy Process in comparison with Best-Worst-Scaling and rating cards: Methodological aspects and ranking tasks

Background: Identifying patient priorities and preference measurements have gained importance as patients claim a more active role in health care decision making. Due to the variety of existing methods, it is challenging to define an appropriate method for each decision problem. This study demonstrates the impact of the non-standardized Analytic Hierarchy Process (AHP) method on priorities, and compares it with Best-Worst-Scaling (BWS) and ranking card methods. Methods: We investigated AHP results for different Consistency Ratio (CR) thresholds, aggregation methods, and sensitivity analyses. We also compared criteria rankings of AHP with BWS and ranking cards results by Kendall's tau b. Results: The sample for our decision analysis consisted of 39 patients with rare diseases and mean age of 53.82 years. The mean weights of the two groups of CR ≤ 0.1 and CR ≤ 0.2 did not differ significantly. For the aggregation by individual priority (AIP) method, the CR was higher than for aggregation by individual judgment (AIJ). In contrast, the weights of AIJ were similar compared to AIP, but some criteria's rankings differed. Weights aggregated by geometric mean, median, and mean showed deviating results and rank reversals. Sensitivity analyses showed instable rankings. Moderate to high correlations between the rankings resulting from AHP and BWS. Limitations: Limitations were the small sample size and the heterogeneity of the patients with different rare diseases. Conclusion: In the AHP method, the number of included patients is associated with the threshold of the CR and choice of the aggregation method, whereas both directions of influence could be demonstrated. Therefore, it is important to implement standards for the AHP method. The choice of method should depend on the trade-off between the burden for participants and possibilities for analyses.