Funnel plot

from Wikipedia, the free encyclopedia
Example of a funnel plot

In the meta-research one is funnel plot , even funnel graphic or funnel chart is a graph that allows a suspicion of publication bias in a meta-analysis to check.

The usual representation is a scatter plot in a Cartesian coordinate system , with the treatment effect plotted on the x-axis against the study size on the y-axis. A funnel plot can be evaluated simply by looking at it: A symmetrical shape is created from a balanced study publication that depicts the results within the scope of natural statistical spread . Therefore, larger studies should achieve more precise results that are closer to the mean of all study results. If the picture is asymmetrical on one side of the mean due to many small studies, this speaks strongly in favor of unpublished or unperformed studies with opposite results.

Criticisms

A major problem with funnel plots is that, being a very simple tool, they do not capture the quality of the included studies. If studies with higher precision actually produce different results than those with lower precision, this would give the wrong impression of publication bias. These criteria must therefore be checked by the analyzer himself. It can also happen that publications on a topic cannot be reduced to a single number because, for example, the choice of test subjects (gender, age structure, ...) or other circumstances can vary.

See also

Web links

  • MIX 2.0 - software to create funnel plots with Excel

Individual evidence

  1. ↑ Weckmann G, Chenot JF, Reber KC: Reading and interpreting meta-analyzes: a practical guide . In: Journal of General Practice . 91, No. 11, 2015, pp. 469–473.  ( Page no longer available , search in web archives )  Info: The link was automatically marked as defective. Please check the link according to the instructions and then remove this notice. @1@ 2Template: Toter Link / www.online-zfa.de  
  2. ^ Matthias Egger, George Davey Smith, M. Schneider & C. Minder: Bias in meta-analysis detected by a simple, graphical test . In: BMJ . 315, No. 7109, 1997, pp. 629-624. PMID 9310563 . PMC 2127453 (free full text).
  3. ^ Joseph Lau, John PA Ioannidis, Norma Terrin, Christopher H. Schmid & Ingram Olkin: The case of the misleading funnel plot . In: BMJ . 333, No. 7568, September 2006, pp. 597-590. doi : 10.1136 / bmj.333.7568.597 . PMID 16974018 . PMC 1570006 (free full text).