A full Bayesian approach to decision & policy making: two cases from research on neuropathic cancer pain

Bayesian methods are useful tools to synthesize research findings from different sources, but today the potential for the use of informative priors is still not well recognized. Neuropathic pain is an emerging research area in palliative care research: an exhaustive description of this phenomenon from published literature is not available yet. In order to inform evidence about the risk of accidental falls in the elderly due to the use of antiepileptics (a medication frequently prescribed for neuropathic pain) and the prevalence of neuropathic cancer pain, an opinion from expert physicians was elicited using structured questionnaires. The opinions about the increased risk of falling were subsequently translated into priors suitable for the introduction into a random-effects Bayesian meta-analysis, combining relative risks from published papers and from experts. A Bayesian hierarchical model can be a feasible method as well to obtain an estimate of the neuropathic pain prevalence according either to literature data and experts opinion. The updated estimates for the risk of falling and for the prevalence, both expressed as point estimates with a credibility interval, represent the synthesis of different sources of information. This is an example of how Bayesian methods can inform decision and policy making when data from the literature are scarce and/or inconsistent.

DEANDREA Silvia; 

2015-02-24

Yeditepe University

JRC71980

http://marc.yeditepe.edu.tr/yircobs12_abstracts_120618.pdf,    https://publications.jrc.ec.europa.eu/repository/handle/JRC71980,