Systematic reviews for the evaluation of interventions that include non-randomized studies

Methodological considerations

Abstract

Systematic reviews to evaluate interventions including the evidence provided by non-randomized studies (NRS), can be very useful to support clinical decisions in subjects that are not well covered by randomized clinical trials (RCTs), such as the evaluation of the effects of long-term treatment, its adverse effects or the evaluation of public health or organizational interventions. The development of these RSLs requires a series of methodological adjustments for the search and selection of the primary studies, the evaluation of their quality and the combination of the evidence. Special care must be taken to identify and appropriately manage the increased risk of bias, especially those derived from inadequate control of confounding factors. This article presents the particularities to take into account when performing RSL to evaluate interventions that include NRS, the tools that have been developed to facilitate their development, and the important considerations to report and discuss the findings. The development of this type of RSL requires a high level of experience by the developer group and a greater amount of economic resources and time. However, when they have been well designed, conducted and analyzed, they can be very useful in clinical decision-making, and can be integrated into the development of the recommendations presented in the clinical practice guidelines