Next generation systematic review methodology

Abstract

Systematic reviews and meta-analyses are widely used to inform guidelines, policy, and practice. Yet, there are several limitations associated with traditional systematic reviews. Potential sources of bias, such as publication bias and selective outcome reporting, can produce misleading results, and when individual studies collect different outcomes or use different measures to assess the same outcomes, this can make them difficult and sometimes impossible to synthesise. Traditional aggregate data meta-analyses give estimates about average effects, but provide limited reliable information on whether intervention effects vary across different populations, or whether differences between intervention characteristics may lead to differential effects. This is particularly problematic in an era that is steering away from a one-size-fits-all approach and toward precision medicine. In addition, traditional meta-analyses only include head-to-head comparisons of two interventions at a time when in reality, there are often more than two options that practitioners need to choose between. To explore these limitations and propose solutions, this thesis presents a series of nine manuscripts.