Barthold D, Brah AT, Hauber B, Simoni JM, Graham SM. The systematic development of attributes and levels for a discrete choice experiment of HIV patient preferences for long-acting antiretroviral therapies. Poster presented at the 42nd Annual North American Virtual Meeting; October 2020.

PURPOSE: Discrete choice experiments (DCE) are widely used in patient preference research. HIV patients have had limited choices for antiretroviral therapy (ART) beyond daily oral pills, but new long-acting products (LA-ART) are in development. Patient preferences for these therapies, elicited via DCE, could help guide development. We describe the systematic process for identifying attributes and levels for a DCE of acceptability of future LA-ART options in the US.

METHODS: Our iterative approach to determining attributes and levels was guided by the four-stage model proposed by Helter and Boehler (2016): data collection, data reduction, removing inappropriate attributes, and wording of attributes. We started with 6 attributes identified in relevant literature, including our own preliminary research. Next, we conducted 11 key informant interviews (digitally recorded and transcribed) following an evolving interview guide. Our team iteratively updated the list of attributes and plausible levels for each attribute. Additionally, we removed irrelevant attributes and specified restrictions on combinations of attribute levels. Finally, we developed an unlabeled optimal experimental design using Ngene software, with a modified Federov algorithm and D-error minimization for at least 3 minutes.

RESULTS: Despite uncertainty about which products are most likely to be developed, key informant discussions converged on 4 delivery modes (long-acting oral, injection under the skin, injection into the muscle, implant) and 6 key attributes. LA infusions and patches were not considered feasible. Treatment effectiveness was dropped because it is a prerequisite for approval and thus the same for all modes. Oral lead-in therapy was split into two attributes (for allergy testing, for viral suppression), whereas injection site pain and reaction were combined into a single attribute. For 4 modes of LA-ART delivery, we reached the following attributes and levels: frequency of dosing (from once a week to once a year), location of treatment (home, pharmacy, clinic), injection site pain (none, mild, moderate), need to check for allergies (not needed, needed), need for oral lead-in to ensure viral suppression before starting LA-ART (from none to 6 months), and late dose leeway (short, long). The final experimental design has 64 choice sets and a D-error of 0.053978.

CONCLUSIONS: We capture the elements that are salient to the acceptability of LA-ART alternatives by patients, and we will confirm these findings through pilot tests of the DCE.

Share on: