Bogdan Grigore

Abstract

The elicitation of probability distributions from experts can be used as a
way to characterise parameter uncertainty in decision models, when evidence is
scarce. This approach has been used routinely for a long time in fields like
ecology or risk assessment, but it can also be helpful in decision models built as
part of health technology assessments.
Eliciting probability distributions from experts is a complex process, and
many aspects of the elicitation process may influence the outcome. The general
literature on elicitation of probability distributions proposes many elicitation
methods, but at the same time, suggests that different approaches may lead to
different results, without defining a clear way of choosing among these
methods. In the context of health technology assessment, elicitation of
probability distributions is an even rarer topic and there is almost no guidance
on what is achievable or what the methodology to achieve it is.
The aim of this PhD thesis was to explore elicitation of probability
distributions to estimate parameter uncertainty in decision models developed as
part of health technology assessments from a practical point of view, that
considered potential methodological trade-offs between minimising the risk of
bias and minimising the resources used for elicitation.
Four studies were conducted as part of this project:
1. A systematic review of current use of elicitation of probability
distributions reported in a health technology assessment context.
2. A comparative study of two elicitation methods used to estimate
parameter uncertainty in a health economic decision model.
3. A comparative study exploring the feasibility of conducting
elicitation remotely.
4. A study exploring five different approaches to weighting individual
experts when mathematically aggregating probability distributions from several
experts.
This thesis covers the findings of these four studies, as well as attempts
to integrate this research in the greater context of health technology
assessment. By wider dissemination of methodological and practical issues of
elicitation in health technology assessment, they could be more clearly
analysed in further research. Furthermore, this will help to increased confidence
in the feasibility of conducting elicitation in the context of HTA.

Document Type

Thesis

Publication Date

2015

Embargo Period

2024-07-19

Download

Share

COinS