Elicitation of expert opinion on uncertain quantities to inform decision models: Negative Pressure Wound Therapy in the treatment of severe pressure ulcers

Abstract

Data elicited directly from experts has great potential value for health technology assessment. However, elicitation exercises designed specifically to collect data for decision model parameters are limited. There are a number of important issues to consider when designing such exercises, especially the need to collect complex data (i.e. conditional probabilities) whilst maintaining the comprehensiveness of the exercise for the expert.

There are no accepted guidelines around the design and conduct of elicitation to inform decision model parameters, only a number of guiding principles that can be used.

This presentation describes the design of a computer-based elicitation exercise undertaken as part of a study to estimate the cost-effectiveness and value of information of Negative Pressure Wound Therapy in the treatment of severe pressure ulcers. Limited, or no, published data was available to inform many model parameters, thus the elicitation aimed to generate estimates of multiple uncertain parameters and to validate analytical assumptions. The elicitation exercise was programmed using Visual Basic for Microsoft Excel and was completed individually by 28 experts in a single session. I will give an overview of the exercise, focussing particularly in issues regarding its design.