Meta-analysis

Regarding meta-analysis, which one of the following statements is TRUE?

a) Is analagous to a systematic review

b) The size of a ‘blob’ in a ‘blobbogram’ reflects the degree of significance found in the individual study

c) If the centre line is crossed by the confidence interval of the combined result,

there is no association between the variables

d) The ‘x’ axis of the results graph is usually expressed as relative risk

e) The funnel plot helps to identify selection bias

Answer: E

Explanation

Meta-analysis continues to suffer from criticism but remains a cornerstone of evidence based medicine. As such, a good working knowledge of how meta-analysis works is essential. Systematic review is a rigorous technique for examining the medical literature. It usually starts with asking a question; working out which database(s) to examine; performing a search; selecting and then reviewing the best evidence; and synthesising an answer to the question with a level of recommendation. Meta-analysis is a statistical technique for combining the results of small trials to provide power through strength of numbers allowing questions to be answered. All meta-analyses start with a systematic review, but not all systematic reviews involve meta-analysis. The size of the ‘blob’ in a ‘blobbogram’ indicates theweighting towards the overall analysis that has been assigned to an individual study. It has nothing to do with the findings in the individual study, just the study’s contribution to the meta-analysis. If the centre line is crossed by the confidence interval of the combined result, there may still be an association but the metaanalysis has failed to demonstrate it. The ‘x’ axis is usually either the odds ratio or favours treatment/favours control. Typically in meta-analysis, selection bias involves the study team missing small studies with negative findings as these are often less likely to be published. Big studies will tend to sit closer to the truemean,whereas smaller studies will spread out further from the mean in both directions. A funnel plot involves the studies plotted with decreasing size up the ‘y’ axis and demonstrated effect on the ‘x’ axis. They will show asymmetry at the broader part of the funnel if such a selection bias exists.

Reference

Crombie IK, Davies HTO. What is meta-analysis? 2nd edn. What is . . .? series.Hayward
Medical Communications, 2009. Online at www.medicine.ox.ac.uk/bandolier/painres/
download/whatis/Meta-An.pdf (Accessed 30 October 2009)