FRCS Revision

This site is private

Enter the password to open the revision library.

Personal revision library · not clinical advice
FRCSRevision
Home / Statistics / Systematic review and meta-analysis
Statistics

Systematic review and meta-analysis

Forrest plots and inverted funnel plots in meta-analysis.

12 questions 2 source pages

Images appear with the first question taken from each source page — tap a question to open it.

12 questions
Q1What is a forest plot?▸
  • Numerical and graphical representation of a meta-analysis
  • Displays individual study results together with the combined overall effect
Q2What is a meta-analysis and how does it differ from a systematic review?▸
  • Meta-analysis = quantitative systematic review combining results of relevant studies to produce an estimation of the overall effect of interest
  • Systematic review = overview of primary studies using explicit and reproducible methods
Q3What information is shown in the columns of a forest plot?▸
  • 1st column: name of study in alphabetical order
  • 2nd-5th columns: incidence of event in control and experimental group
  • 6th column: weight - reflects precision of the study (size and CI)
  • 7th column: RR and CI interval
Q4What do the symbols in a forest plot represent?▸
  • Square = weight of study
  • Horizontal line = confidence interval
  • Central line = no-effect line; if the CI crosses it the result is not statistically significant
  • Diamond at the bottom = combined effect of the studies
Q5How is heterogeneity assessed and graded in a meta-analysis?▸
  • Chi-squared heterogeneity test (Q test) or I2
  • I2 <50 = low, 50-75% = moderate, >75% = high heterogeneity
  • I2 = the proportion of variation that is due to heterogeneity rather than chance
  • p<0.05 rejects the null hypothesis of no statistical heterogeneity
Q6What do 'fixed' and 'M-H' mean and how is heterogeneity managed?▸
  • Mantel-Haenszel (M-H) test used to combine studies (Review Manager, Cochrane Library)
  • Fixed-effects model: one true effect underlies all studies
  • Random-effects model: normal distribution of true effect sizes; mean and variance estimated
  • Management: subgroup analysis, random-effects model (may report plot without combined estimate), meta-regression for confounders
Q7What is heterogeneity per the lecture definition?▸
  • Variability in patient characteristics within each study
Q8What is meta-regression and what is it used for?▸
  • An extension of subgroup analysis allowing the effects of multiple factors to be investigated simultaneously
  • Used to look for confounders
  • Similar to simple regression: an outcome variable predicted from one or more independent variables
Q9What is an inverted funnel plot?▸
  • Scatter plot of the intervention effect estimates from individual studies
  • Plotted against some measure of each study's size or precision
Q10What is an inverted funnel plot used for?▸
  • Helps detect publication bias in a meta-analysis
  • Small size study with lower methodology quality tends to overestimate true effect
Q11What are the axes and reference lines of an inverted funnel plot?▸
  • Y axis: precision of study (standard error, or standard deviation/square of sample size)
  • X axis: effect of treatment (relative risk)
  • Vertical dotted line: estimated combined RR of the meta-analysis
  • Diagonal dotted lines form the funnel
Q12How should studies be distributed and what does asymmetry indicate?▸
  • Upper dots: larger sample size/more precise, lie closer to the vertical dotted line
  • Lower dots: smaller/less precise studies, more scattered and symmetrically distributed
  • Asymmetry suggests publication bias, or small low-quality studies overestimating the true effect
← Data types, outcome measures and epidemiOsteosarcoma case - assessment and stagi →