Abstract
Systematic reviews can provide timely, reliable evidence on which to make informed decisions. In order to make balanced decisions, information is not only needed on the benefits of an intervention, but also on its adverse effects. Yet few systematic reviews incorporate adverse effects data in their analysis. There is currently a lack of guidance on how to identify adverse effects data, this may impede systematic reviewers. This thesis helps address this situation by evaluating and optimising the methods for retrieval of research evidence for systematic reviews of adverse effects.
The first stage of this programme of research critically reviews the methodological literature relating to the retrieval and inclusion of adverse effects data, including aspects such as the impact of study design (for example RCTs and cohort studies), database search strategies (for example in MEDLINE and EMBASE), sources of data (including database and non-database sources), publication status and funding status.
Second, the results of a survey of the literature searching methods used in 849 systematic reviews of adverse effects are presented. Data were collated on aspects such as sources searched, search strategy design and the standard of reporting of the methods used. The reviews are published over a 17 year time period (1994-2011) thus enabling time trends analysis. The methods used in these systematic reviews of adverse effects are also compared with those reported in surveys of other types of reviews.
Further potentially relevant evidence is incorporated to address gaps identified in the literature. A detailed analysis is provided of the contribution of different sources of data for adverse drug reactions using 58 included studies from a case study systematic review. The same case study systematic review is then used to measure the performance of adverse effects search filters in MEDLINE and EMBASE.
Finally 242 included papers from a series of 26 systematic reviews are evaluated to strengthen the evidence base regarding adverse effects search filters and to assess individual adverse effects search terms in MEDLINE, EMBASE, and Science Citation Index (SCI).
The strengths and weaknesses of the analyses are discussed and implications for practice and guidance presented along with recommendations for future research.
