posted on 2015-03-06, 16:44authored byFelix Adjuah Achana
This thesis focuses on the challenges of evidence synthesis to inform healthcare decision
making within public health. It encompasses both methodological advancement and practical
application of existing synthesis methodology, using as an example - accidents prevention in
children to illustrate application of the methods within a public health context.
The thesis commences with a systematic review of NICE public health appraisals to identify
the barriers to quantitative synthesis of evidence in public health. Then focusing on the
prevention of unintentional poisonings in pre-school children, a series of network metaanalyses
of the effectiveness evidence are conducted, demonstrating how complex synthesis
methodology can be employed to help overcome some challenges of evidence synthesis in a
identified in the review of the NICE public health appraisals.
New synthesis methodology is then developed in which the standard network meta-analysis
model is first extended to include a covariate for the baseline risk and then to a multiple
outcome settings. Baseline risk is a proxy for unmeasured but important patient-level
characteristics, which may be modifiers of the treatment effect in a meta-analysis. Thus
adjusting for it can account for heterogeneity across different study populations and identify
those more likely to benefit from the intervention. The multiple outcome models account for
the dependency structure within the data which is important in a decision modelling context,
as correlations between effect estimates on different outcomes may have implications for
estimating the net benefit associated with treatment.
Finally, a substantive decision analytic model is presented incorporating results from the
network meta-analysis and application of the methodology developed to the poison
prevention data. The analyses suggest that compared to usual care, more intensive home
safety interventions are more effective in preventing medicinal poisonings in pre-school
children but are unlikely to be cost-effective for the UK NHS unless policy makers are
willing to pay upwards of £75,000 for every QALY gained.