Meta-analysis is a statistical procedure that integrates the results of several independent studies considered to be “combinable “. It is a way of structuring the process through which a thorough review of previous research is carried out. It should be free from bias and if conducted well allows a more objective appraisal of the evidence than traditional narrative reviews seen in the past. Synergy between studies and more precise estimates of treatment effect can often be evidenced.
Start with the preparation of a detailed protocol. Look at different statistical methods used to combine data and select one.
Remember to standardise the outcome measures. Define the end point e.g. if it is binary “ disease” versus “no disease”. Consider the odds ratio and risk. Remember that small studies are more subject to chance and should be given less weight. If you have lots of small studies it may not be suitable to combine the results for this reason. Graphic displays illustrate the results well and should be included, remember to label the axis and define the scale and unit of measure.
After conducting the analysis ask yourself does the conclusion reached make sense? How robust are they? Do they make clinical sense and do they contribute to the rational decisions about the management of patients and improve patient outcomes ?