OBJECTIVES:

A stated-preference study aimed to quantify patients’ benefit-risk preferences for alternative devices used in revascularization procedures for peripheral arterial disease (PAD). A main objective was testing whether parallel discrete-choice experiment (DCE) and threshold-technique (TT) data would yield similar maximum acceptable risk (MAR) estimates.

METHODS:

Eight DCE questions offered patients a choice between 2 device options for a revascularization procedure for PAD. Four probabilistic attributes included the chance of a repeat revascularization procedure at 2 years (10% or 30%) and 5 years (30%, 40%, or 50%) and mortality risk at 2 (2% or 5%) and 5 years (8%, 10%, 16%, 20%). The key metric of interest was the maximum 5-year mortality risk patients would accept in excess of 8%, for a device that could reduce the risk of another procedure by 2 years from 30% to 10%, and by 5 years from 40% to 30%. The TT offered patients a fixed level of improvement with 5-year mortality risk starting at 8% and increasing by 2 percentage points up to 20%. DCE and TT order was randomized. DCE data were analyzed with mixed-logit models and TT with interval regressions. MARs were calculated for the sample and each patient. Participants were recruited from seven U.S. medical centers.

RESULTS:

Among those who saw the DCE first (n=136), the mean MAR was 13.26% [95% CI: 12.20, 14.31]. Among those who saw the TT first (n=134), the mean MAR was 14.00% [12.22, 14.78]. Sensitivity analysis of MARs to order shown and risk presentation were similar [DCE: 11.94-13.71, TT: 12.84-14.14]. At the individual level, discordance between DCE and TT ranged from 0 to 12 percentage-points with a mean difference of 2.5.

CONCLUSIONS:

DCE and TT preference methods yielded similar mean MAR estimates at the sample level. However, concordance at the individual level varied widely.