OBJECTIVES: Given there is no consensus about the best way to elicit patients’ preferences, this study empirically test the consistency of responses to two different preference elicitation methods. Such research has been encouraged by the FDA and is one of the key motivators of the Innovative Medicine Initiative funded Patient Preferences in Benefit-Risk Assessments during the Drug Life Cycle project (PREFER).

METHODS: Two methods, thresholding and adapted swing weighting (ASW), were adopted to elicit patients’ minimum acceptable benefit (MAB) for pegvaliase, a treatment for phenylketonuria (PKU). Both elicitation methods used the same attributes and associated levels: the ability to lower phenylalanine (Phe) levels, and the likelihood of experiencing allergic reactions. To enable the reasons for differences in MAB obtained from these two exercises the participants were asked to outline their rational for their choices

RESULTS: Patients would require a minimum likelihood of 22.67% (ASW) to 34.4% (thresholding) of achieving a Phe level less 360 μmol/L before they would tolerate the risks. This is less than the benefit generated by pegvaliase. There was no relationship between MAB and patient’s current Phe level, perception of PKU control, or satisfaction with treatment. Seven participants had their MAB increase significantly between the ASW and the thresholding exercise. In four cases, this was explained by their refusal to take an injectable, which was an aspect highlighted in the introduction to the thresholding but not in the ASW exercise.

CONCLUSIONS: The insight generated by both methods supported patients would be willing to take pegvaliase despite its risks, and both methods drew the same conclusions with regards to treatment preference. Also, this study highlights the impact of the elicitation format and framing of questions on the elicited trade-offs. The introduction of the treatment characteristic could explain most of the discrepancies between the MABs in the two exercises.