Please confirm topic selection

Are you sure you want to trigger topic in your Anconeus AI algorithm?

Updated: Dec 29 2021

Systemic Review and Meta-Analysis

  • Introduction
    • Systemic reviews synthesize previous studies to produce more precise and generalizable data
      • synthesized studies can be interventional or observational
    • Meta-analysis is the statistical procedure for combining numerical results from synthesized studies
  • Systemic Review
    • Systematic and procedural method of collecting, evaluating, and synthesizing qualitative or quantitative literature to answer a question
      • methods and selection criteria must be specified in advance
    • May or may not include meta-analysis
  • Meta-Analysis
    • Statistical method to combine data from multiple studies
      • weighted average
    • Advantages
      • better precision than individual studies
      • improves generalizability of study findings
      • considered to be the highest level of clinical evidence
    • Limitations
      • quality of individual studies
      • bias of individual studies
        • must be assessed and accounted for
      • variability in study methods
        • studies must be similar enough to be meaningfully combined
        • e.g., meta-analysis may not be helpful for a study looking to compile evidence about the effects of reading on depression scores if the primary studies all focused on reading different genres
          • would be more helpful to answer a specific question such as "How does reading fiction influence depression scores?"
      • study heterogeneity
        • when variation in effect size of included studies is greater than would be expected by pure chance
          • e.g. one study demonstrates a massive treatment effect and another demonstrates a negligible one
          • could be due to differences in participant characteristics
            • treatment could have larger impact in certain populations and pooling the results without examining that possibility will obfuscate that
          • statistical concerns
            • high heterogeneity could suggest absence of "true" effect, rendering pooled data meaningless
      • subject to publication bias
        • outcome and statistical significance of a study or experiment influences the likelihood of its publication
        • more papers with positive results are published despite similar quality
        • when present, sampled publications are not a true representation of gathered evidence
        • may lead to increase in false conclusions
  • Funnel Plot
    • Evaluates for the presence of publication bias
    • Used in systemic reviews and meta-analyses
    • Plots precision versus results measure (e.g. odds ratio, mean, and relative risk) of included studies
    • Should take on a funnel (triangle) shape
      • studies with high precision (high value on y-axis) should fall near the average result measure (center of x-axis)
      • studies with low precision should fall scattered on either side of the average result
    • Deviation from funnel shape indicates publication bias
  • Forest Plot
    • Used in systemic reviews and meta-analyses to visually display results from individual studies
    • Vertical "line of null effect" at point on horizontal axis representing no association between exposure and outcome
    • Positive and negative results measures (e.g., odds ratio, mean, and relative risk) of included studies plotted as boxes or shapes on either side of the line
      • relative size of the shape indicates relative size of the study
      • 95% confidence interval (CI) plotted as a line extending horizontally from each point
        • any CI crossing the vertical line indicates the results are statistically insignificant
      • Combined point estimate (vertical line) and confidence intervals represented by a diamond
1 of 0
Private Note

Attach Treatment Poll
Treatment poll is required to gain more useful feedback from members.
Please enter Question Text
Please enter at least 2 unique options
Please enter at least 2 unique options
Please enter at least 2 unique options