Quality assessment of healthcare intervention studies in systematic reviews: improving established tools and approaches for context-specific assessments

Systematic reviews play an essential role in informing healthcare decision-making. A very important part of the systematic review process is an assessment of the quality of the included studies. This allows differentiation between studies which are methodologically robust and reliable, from studies which may have design weaknesses. Although quality assessment tools currently exist which are established and widely used, they nevertheless continue to develop and evolve. The main aim of the publications in this thesis is to enhance quality assessment outputs. This should lead to improvements in the value and reliability of systematic review results and conclusions on treatment efficacy and safety. Some papers in the thesis propose improvements to established quality assessment tools: the Cochrane risk of bias tool for assessing randomised trials and QUADAS-2 for diagnostic accuracy studies. These ideas have had an impact. For example, evaluation of patient baseline characteristics has recently been incorporated into the next version of the Cochrane risk of bias tool to improve evaluations of randomisation process biases. Other papers relate to areas of healthcare research where a context-specific approach to quality assessment is needed i.e. where reliance solely on established tools may result in important issues being missed. These papers illustrate the importance of reviewers being aware of the range of issues which might affect the applicability (relevance) of a trial’s results to other patient groups and settings. Enhanced quality assessment outputs should provide systematic reviewers with a better understanding of study design issues. This should optimise systematic review recommendations for future research, by providing details of the important design issues which need to be addressed, together with suggestions for the most appropriate and viable study designs for future studies to consider.