Does It Look like Evidence? Assessing the Stated-Preference Evidence Base for Inflammatory Bowel Disease

OBJECTIVES:

The maturation of health-preference research is indicated by the large number of published studies that have accumulated in some therapeutic areas. It is now possible to begin thinking of preference data in terms of evidence bases, similar to clinical data. We undertook a proof-of-concept study to assess the feasibility of identifying consensus risk-tolerance estimates from the body of evidence available on treatment preferences for Crohn’s disease and ulcerative colitis (IBD).

METHODS:

We identified 22 published IBD preference studies, 7 of which reported discrete-choice-experiment (DCE) estimates useable for calculating maximum acceptable risk (MAR). Consensus published estimates were obtained by regressing MAR estimates on study-design characteristics. We also have obtained access to 5 original DCE datasets. The original data were pooled to estimate a serious-infection-scaled, MAR-space, data-fusion model. Assumptions were required to harmonize attribute definitions.

RESULTS:

Published results provided 314 individual MAR estimates. Including serious-infection increased the MAR for given efficacy by 0.9% while including malignancy decreased it by 0.8%. The pooled original datasets contained a total of 1,366 respondents and over 21,000 choices. A treatment that improved IBD symptoms from moderate to remission, but with an annual cancer risk of 1%, had an average annual serious-infection equivalent risk tolerance of 16.7%.

CONCLUSIONS:

Stated-preference evidence bases in well-studied therapeutic areas can help establish consensus values for risk-tolerance measures, increase credibility for using stated-preference data to inform regulatory and clinical decision making, and enable leveraging previous research for benefit transfers to provide values in in the absence of sufficient time and funding for original studies, as well as help inform efficient, targeted new studies to fill identified gaps in the existing literature.