Meta-analysis as a source of evidence in gastroenterology: a critical approach.

Meta-analysis is increasingly used in hepatogastroenterology. Meta-analysis is of value to provide a systematic review of related trials and to display their results in an objective, easily understandable manner. When the trials are sufficiently homogeneous, meta-analysis can document the superiority, (a), or the lack of superiority (b) of a treatment with respect to another (e.g., (a) Interferon plus ribavirin vs Interferon for chronic hepatitis; (b) 5-ASA vs sulfasalazine for maintaining remission in ulcerative colitis). However the interpretation of meta-analysis requires caution. Meta-analysis can be unreliable or unstable if based on a few, small trials (e.g., Tamoxifen vs non-active treatment for hepatocellular carcinoma), or if distorted by confounding variables and publication bias (e.g., glucocorticoids vs standard treatment in alcoholic hepatitis). Eventually, qualitative heterogeneity makes the pooled results of meta-analysis meaningless or questionable (e.g., endoscopic sclerotherapy for prevention of first variceal bleeding in cirrhosis) and should prompt the search for its sources to plan future studies. Finally, meta-analysis of trials measuring the treatment effect of a drug vs a placebo when an active drug is available for comparison provides the limited informative content for the physician of the individual trials (e.g. 5-ASA vs placebo for maintaining remission in ulcerative colitis).