Systematic reviews and meta-analysis

Abstract

Systematic reviews play an important role in healthcare decision-making. When conducted correctly, they provide up-to-date, comprehensive, and replicable summaries of evidence. Authors of systematic reviews are expected to develop a protocol that outlines the research question and key methodological features of their review. A comprehensive and exhaustive search should be conducted, followed by screening to capture studies that meet the prespecified inclusion criteria. Once the relevant studies have been identified, data will be extracted, using a dedicated tool that permits the review authors to confirm the eligibility of the study and collect information on its design, risk of bias, and results. Sufficiently similar data may be pooled using meta-analytic techniques or synthesized narratively. A summary of the overall quality of evidence for each outcome is an essential component of a systematic review. The main concerns with systematic reviews are (1) selection bias: systematic exclusion of relevant studies due to publication status or language; (2) indexing bias: failure to identify relevant studies because they are not indexed accurately; and (3) information bias: missing or inaccurate information in the included studies. Other approaches to evidence synthesis include mapping the evidence with scoping reviews; conducting overviews of systematic reviews; using individual patient data; conducting network meta-analyses for multiple comparisons; conducting rapid reviews when evidence is needed urgently; synthesis of diagnostic accuracy data; and synthesis of qualitative data. Systematic reviews often inform clinical guidelines and require careful planning and execution by teams with content and methodological expertise.