Development of evidence synthesis methods for health policy decision making - A chain of evidence approach
thesisposted on 2010-02-03, 12:52 authored by Clare Louise Gillies
This project comprises a critical exploration and development of methods for the synthesis of evidence, using a chain of evidence approach, from diverse, yet inter-related, sources. The methodologies were explored through the development of a comprehensive decision model to assess different health policies in respect to screening for type 2 diabetes mellitus (T2DM). Four strategies were compared which were, no screening (current policy), screening for T2DM alone allowing for early diagnosis and treatment of the condition, and two strategies whereby both impaired glucose tolerance (IGT) and T2DM were screened for, allowing for early treatment of T2DM and for either lifestyle or pharmacological interventions to be applied to those with IGT in an attempt to delay the onset of T2DM. The comprehensive decision model developed here was innovative when compared to current published models in a number of ways. Firstly the entire model was encompassed within a single flexible framework, which has a number of advantages, and secondly as much of the available data as was feasible to use, was incorporated into the model inputs. A number of methodological issues and techniques were explored during the development of the comprehensive decision model. These included mixed treatment comparison analyses, assessment of baseline risk on intervention effects and the use of individual patient data. A number of sensitivity analyses and model extensions were carried out to assess the parameters with most influence on model results, and to adapt the model to different screening scenarios. The results of the model provide evidence that a screening strategy for IGT and T2DM, followed by appropriate treatment and interventions appears to be a cost-effective screening strategy. Uncertainty still surrounds the cost-effectiveness of screening for T2DM alone and further research is required. Running decision models within a Bayesian, comprehensive decision modelling framework, allows for model flexibility and has advantages over more conventional modelling techniques.
Date of award2007-01-01
Awarding institutionUniversity of Leicester