Bias in meta-analysis and funnel plot asymmetry.

International experiences reveal the important role played by scientific research and systematic study a problems, in effectively tackling change in the health sector. Meta-analysis was introduced to address the problem of synthesizing the large quantity of information on a particular subject. It is viewed, only as a step in the process of developing better tools to quantify information across studies. The selection of trials for inclusion in a meta-analysis may be biased if selection is restricted to published trials, to trials published in English language journals, to trials published in prestigious journals or to trials cited by other authors. Funnel plot is graphical display of sample size plotted against effect size for the studies included in a meta-analysis, which can be used to investigate bias. When all the studies have been located, the distribution of points should resemble a funnel. If there are gaps in the funnel shape it indicates that some studies may have not been published or located. In evaluating bias, we use meta-analysis studies about radiotherapy alone versus combined radiotherapy and chemotherapy in stages IIIa and IIIb non-small cell lung cancer. A simple analysis of funnel plots provides useful test for the likely presence of bias in meta-analyses, but as the capacity to detect bias will be limited when meta-analyses are based on a limited number of small trials the results from such analyses should be treated with considerable caution.