25. Meta-analysis, systematic review

From greek.doctor
  • These are the highest-quality studies
  • No new data is collected; results from previous studies are combined
  • Combining multiple studies reduces biases and improve statistical power
  • Meta-analysis
    • Data from multiple similar studies is processed with statistical methods
    • Can be performed when there are multiple studies addressing the same question
    • Answers questions like:
      • What is the size of the effect?
      • Is the effect consistent across studies?
      • What is the strength of evidence for the effect?
    • Steps
      • Identify relevant studies and exclude poorly conducted studies
      • Perform statistical analysis
    • Which studies are included and which are excluded are very important for the validity of the results
  • Systematic review
    • Data from multiple similar studies is evaluated, but not necessarily processed with statistical methods
      • Systematic review often include a meta-analysis
    • Systematic review is needed because:
      • There are way too many individual studies to keep in mind
      • The sheer number of studies isn’t wanted; what is wanted is high-quality evidence
      • Individual studies contain bias, which can be reduced by systematic review
      • It produces a clear and consistent summary of the science on the question
      • Often times different studies yield conflicting results, which a systematic review may yield information on
    • May include unpublished data, to prevent publication bias
    • Steps
      • Defining an appropriate question
      • Identify relevant studies and exclude poorly conducted studies
      • Assess quality of each study or report
      • Combine findings from the studies
      • Interpret findings and present a balanced and impartial summary